91
Consultation Paper Draft Extended Reserve Selection Methodology Publication date 11 October 2016 Submissions close: 5pm on Tuesday 29 November 2016 (updated)

Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

  • Upload
    others

  • View
    1

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

Consultation Paper

Draft Extended Reserve Selection Methodology

Publication date 11 October 2016

Submissions close: 5pm on Tuesday 29 November 2016 (updated)

Page 2: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 2 of 91

Contents

Executive summary ................................................................................................................ 4

1 Introduction ............................................................................................................. 6

1.1 Purpose of this paper ................................................................................................ 6

1.2 The affected parties .................................................................................................. 6

2 Process and timeline .............................................................................................. 7

2.1 How to make a submission ....................................................................................... 7

2.2 Timeline and next steps ............................................................................................ 7

3 Requirements .......................................................................................................... 8

3.1 Operational requirements .......................................................................................... 8

4 Structure of the methodology ................................................................................ 9

5 Description of the methodology ............................................................................ 10

5.1 Key features of the methodology ............................................................................... 10

5.1.1 Use of demand units as the basis of efficient procurement ....................................... 10

5.1.2 Use of historic information as a proxy for future performance.................................... 10

5.1.3 A consistent, accurate dataset is essential ................................................................ 12

5.1.4 Improved transparency ............................................................................................. 12

5.1.5 Payments for extended reserve ................................................................................ 12

5.2 Overview of the selection process ............................................................................. 13

5.2.1 Inputs to the selection process .................................................................................. 13

5.2.2 List of constraints and parameters ............................................................................ 13

5.2.3 Selection process description.................................................................................... 14

5.3 Example procurement schedule ................................................................................ 16

5.3.1 Example procurement schedule analysis .................................................................. 16

5.3.2 Warning – explanatory note ...................................................................................... 16

5.4 The proposed flexible solve process ......................................................................... 17

5.5 The proposed methodology for a limited selection process ....................................... 19

6 Consultation on aspects of the methodology ....................................................... 20

6.1 AUFLS provision costs .............................................................................................. 20

6.1.1 The cost calculations ................................................................................................ 20

6.1.2 Use of standard values is proposed .......................................................................... 22

6.1.3 The proposed set of standard costs .......................................................................... 22

6.1.4 An ‘implementation cost’ is not applied ..................................................................... 26

6.1.5 The ‘relay fast response incentive’ is not applied ...................................................... 27

6.1.6 The relay flexible service cost is not applied – is it an incentive? .............................. 27

6.1.7 The costs are intended to cover technical requirements costs .................................. 28

6.2 Interruption costs and their application ...................................................................... 29

6.2.1 The interruption cost calculation ................................................................................ 29

6.2.2 The interruption costs ............................................................................................... 30

6.2.3 Expected interruption hours (EIH) ............................................................................. 30

6.3 Can the selection process be based more on customer class? ................................. 31

6.3.1 Treatment of public health and safety customers ...................................................... 32

6.4 Demand unit information requirements ..................................................................... 34

6.4.1 Quantity requirement ................................................................................................ 34

6.4.1.1 The 4-year requirement ............................................................................................. 34

Page 3: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 3 of 91

6.4.1.2 Missing years must be estimated by the extended reserve manager ........................ 35

6.4.1.3 The 60% offtake requirement .................................................................................... 36

6.4.2 Information description and the data specification ..................................................... 40

6.4.3 The accuracy requirement ........................................................................................ 41

6.4.3.1 Accuracy definition of the load profile information ..................................................... 41

6.4.3.2 Accuracy definition of the customer type information ................................................ 46

6.4.4 Limitations on asset owners involved ........................................................................ 47

6.5 North Island offtake information (reference dataset) .................................................. 48

6.6 Constraints applied to meet technical requirements .................................................. 48

6.6.1 ‘At all times’ requirement ........................................................................................... 48

6.6.2 Use of penalty costs is minimised ............................................................................. 48

6.6.3 AUFLS block thresholds ............................................................................................ 49

6.6.4 AUFLS target levels .................................................................................................. 51

6.6.5 Fast response preference ......................................................................................... 51

6.6.6 Quality level .............................................................................................................. 52

6.7 Constraints applied to meet other objectives ............................................................. 53

6.7.1 Selection cap of 60% per asset owner ...................................................................... 53

6.7.2 Selection of flexible demand units ............................................................................. 56

6.7.3 Minimum load buffer ................................................................................................. 58

7 Evaluation against the Code principles ................................................................. 59

8 Extended reserve procurement schedule template .............................................. 60

8.1 Publication and commercial sensitivity ...................................................................... 60

8.2 Consultation is technical in nature and proposed to be brief ..................................... 61

9 Operational design and obligations on extended reserve providers .................. 63

9.1 Overview of the operational design ........................................................................... 63

9.2 Management of change over time ............................................................................. 64

9.3 Operational obligations on extended reserve providers ............................................. 65

9.4 Rationale for the quarterly data provision requirement .............................................. 70

9.4.1 The extended reserve manager will need a contiguous dataset ................................ 70

10 Compensation payments for extended reserve .................................................... 75

10.1 The Code and asset owner feedback to date ............................................................ 75

10.2 Options for a compensation payment mechanism ..................................................... 76

10.3 What is the policy intent of having a payment mechanism? ...................................... 77

10.3.1 Can introducing a payment mechanism achieve this policy intent? ........................... 77

10.3.2 Incentives and payments .......................................................................................... 79

10.4 Methodology for calculating the payment .................................................................. 80

Appendix A: Draft Selection Methodology.......................................................................... 81

Appendix B: Example Procurement Schedule ................................................................... 82

Appendix C: Analysis of Example Procurement Schedule ................................................ 83

Appendix D: Data specification ........................................................................................... 88

Appendix E: Draft of Part 6 of Schedule 4 .......................................................................... 89

Appendix F: Questions ........................................................................................................ 91

Page 4: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 4 of 91

Executive summary

This is the consultation paper for the first draft extended reserve selection methodology

(methodology). The methodology is prepared under the Part 8 of the Electricity Industry

Participation Code 2010 (Code). The consultation is aimed at all relevant North Island asset

owners who will be required to submit information into the selection process, and who may

become extended reserve providers.

The extended reserve manager would like to thank all the North Island distribution companies

and directly connected consumers who have participated in the development of the

methodology, for your valuable feedback through workshops and supply of trial data for 4 years

of three quarters of the North Island offtake. Using this information, we provide an Example

Procurement Schedule to illustrate the impact of the methodology.

Comments received on the draft methodology and Example Procurement Schedule from the

system operator and Authority as required by the Code, have been taken into account in this

methodology presented for consultation.

We propose to require a minimum of 60% of offtake from each asset owner for four historic

years, and to select no more than 60% of load from any one asset owner. The methodology

recognises that asset owners are best-placed to use their discretion to withhold demand units

with customers whom they consider not suitable for selection. We also propose that demand

units containing solely public health and safety customers should not be eligible for submission.

The methodology critically relies on accurate and complete information from asset owners,

particularly regarding the half-hourly load profile and customer class information. We define

accuracy in the data specification. We have set out more specific requirements in the data

specification to promote consistent and more accurate customer class allocation, and propose

to require curtailable interruptible load (IL) profiles to be subtracted from the half-hourly load as

the most accurate method to avoid double-counting AUFLS with IL.

The methodology is prepared in accordance with the Code requirements for cost-effective

procurement:

using a realistic set of provision costs as surveyed by Beca in February 2016, with some

slight modifications to their application, and

applying customer interruption costs using the generic values and expected interruption

hours provided by the Authority and asset owner information on the proportions of

customer load on each demand unit.

The methodology aims to meet the system operator’s technical requirements by selecting the

most suitable set of demand units that collectively meet the AUFLS blocks requirements for

least cost. The Example Procurement Schedule achieves a near-perfect performance result

against the historic year AUFLS block thresholds. However, in using historic half-hourly

averaged load profiles as a proxy for future performance, the methodology cannot guarantee to

meet the system operator’s ‘at all times’ requirement.

Page 5: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 5 of 91

The extended reserve manager proposes to procure at least 10% of demand units for standby

reserve, to provide for flexibility during the 5-year operational period.

Additionally we propose that all AUFLS relays are configured and tested for as many AUFLS

block settings as the relay is capable of holding. As we did not get specific feedback on this

proposal in workshops, we specifically seek both more information and your views on this

proposal.

During operations it is proposed that providers submit load profile information on their demand

units every quarter. The extended reserve manager is to review the information and apply the

flexible selection process (in the methodology) to determine whether any standby demand units

need to be armed to maintain good AUFLS block performance.

The annual selection cost (operating and capital, for five years) of the Example Procurement

Schedule is $5.96 million. However, the extended reserve manager cautions that in scenarios

the annual selection cost under this methodology increased to up to $7.8 million, where asset

owners supply only the minimum 60% of offtake and where the larger demand units are not

submitted. The net cost impact on an asset owner will depend on the overall quantity, MW size

and relay characteristics of the demand units that are submitted.

The extended reserve manager proposes to introduce compensation payments for extended

reserve providers, using the same provision and customer interruption costs used for selection,

with the one exception that payments for customer interruption will not be paid for demand units

when they are on standby (as there is no interruption risk to customers while on standby).

Finally, this draft methodology enables the technical improvements to the existing AUFLS

scheme to be realised. It also enables better transparency in AUFLS provision, and a more

efficient selection of demand units for AUFLS service across the North Island.

The extended reserve manager invites feedback on the draft methodology.

Page 6: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 6 of 91

1 Introduction

1.1 Purpose of this paper

The purpose of this paper is to present the first draft extended reserve selection methodology

(methodology) for consultation. The consultation process is conducted in accordance with Part

2 of Schedule 8.5 of the Code.

The methodology has been prepared in accordance with clauses 8.54G and 8.54H of the

Electricity Industry Participation Code 2010 (Code), which with the operational design approach

developed by the Electricity Authority (Authority), constitute the requirements that the

methodology aims to meet when selecting extended reserve.

The methodology is attached as Appendix A. This paper presents and explains key matters in

the methodology to enable affected parties to provide feedback.

1.2 The affected parties

The extended reserve manager must consult with persons that the extended reserve manager

thinks are representative of the interests of persons likely to be substantially affected by the

draft methodology. These affected parties are the primary audience of this paper.

The extended reserve manager considers the affected parties to be:

Asset owners who are potentially extended reserve providers and/or who are potentially

liable to pay for procured extended reserve, and their representative bodies;

Traders and ancillary service agents who may be required to provide information to

asset owners to enable them to meet their information obligations.

The extended reserve manager does not consider other parties to be substantially affected by

this methodology. However, other parties who have an interest are welcome to submit.

Given the intended audience, background information on extended reserve is not provided here.

The reader is invited to review the Authority’s extended reserve Code amendments and the

extended reserve manager’s and system operator’s work on this project, accessible via:

Electricity Authority: http://www.ea.govt.nz/development/work-programme/risk-

management/efficient-procurement-extended-reserve

Extended reserve manager: http://www.nzxgroup.com/who-we-are/business-overview/nzx-

energy/consultations-submissions/extended-reserve

System operator: https://www.transpower.co.nz/system-operator/so-projects/extended-reserve

Page 7: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 7 of 91

2 Process and timeline

2.1 How to make a submission

You are invited to provide feedback to the contents of this document and specifically to the

questions provided in Appendix F. Responses must be emailed to the extended reserve

manager ([email protected]) with ‘Consultation feedback’ in the subject line.

Responses must be received by 5pm on Tuesday 29 November 2016 (this is the new

updated deadline) for full consideration. The extended reserve manager will acknowledge

receipt of all emails within two business working days. If you do not receive an electronic

acknowledgment within two business days please contact Tania Ahmed on 04 498 0057.

Please identify any confidential information. The extended reserve manager may make

available non-confidential aspects of submissions on its website. Submitters should indicate any

documents attached in support of the submission in a covering letter and clearly indicate any

information that is provided on a confidential basis. All information received as part of

submissions will be made available to the Authority and the system operator. Please note that

all information provided to the Authority and system operator is subject to the Official

Information Act 1982.

2.2 Timeline and next steps

The extended reserve manager intends (subject to feedback and Authority approvals) to target

the following dates to complete the selection process:

Event Start End

Draft selection methodology consultation phase 11 October 2016 29 November 2016 (updated)

Final methodology provided to the system operator for comment and to the Authority for approval

20 December 2016

Methodology published following Authority approval 14 February 2017

Data request made on participants under the Code 14 February 2017 14 April 2017

Data analysed, revisions requested 15 April 2017 5 May 2017

Data revisions received 5 May 2017 19 May 2017

Draft procurement schedule reviewed by system operator and Authority

26 May 2017 26 June 2017

Draft procurement schedule consultation phase 27 June 2017 14 July 2017

Authority decides approval of procurement schedule 8 September 2017

Procurement schedule and procurement notices published

11 September 2017 11 September 2017

Page 8: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 8 of 91

3 Requirements

The requirements for the methodology are set out in clauses 8.54G and 8.54H of the Code. In

essence, the extended reserve manager must prepare a draft methodology that:

1. specifies how extended reserve will be procured to meet the requirements set out in the

system operator’s Extended Reserve Technical Requirements Schedule (TRS)

2. is based on principles balancing extended reserve provider and system operator

interests, enabling cost-effective procurement, and balancing certainty and flexibility

3. specifies how the methodology applies to each island

4. identifies who must provide what information by when during a selection process

5. explains the basis on which selection occurs (the actual method of selection)

6. sets default terms and conditions that extended reserve providers must follow when

providing extended reserve, and

7. specifies how payments for extended reserve provision are set.

3.1 Operational requirements

As a service provider to the Authority, the extended reserve manager must provide its services

in accordance with the Authority’s statutory objective to promote competition in, reliable supply

by, and efficient operation of, the electricity industry for the long-term benefit of electricity

consumers.

Over the past year the Authority has led a process to specify the operational approach to be

taken once extended reserve is selected and in service. The extended reserve manager has

worked with the Authority and the system operator to develop this design.

The proposed ‘simplified operational design’ specifies the roles and responsibilities of each

service provider and of extended reserve providers. After being presented in the extended

reserve workshop 3 in April 2016 it has since has been developed in more detail.

The obligations on asset owners selected for extended reserve set out in Schedule 2 (terms and

conditions) of the methodology reflect this operational approach. The approach and obligations

are discussed in the operations section in this paper.

Page 9: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 9 of 91

4 Structure of the methodology

The draft methodology is attached in Appendix 1. It follows the original structure of the indicative

version appended by the Authority to its extended reserve Code consultation in mid-2014. The

table maps the Code and operational requirements to the methodology to aid navigation.

Table 4-1: Methodology structure mapped to Code requirements

Requirement Component Methodology reference Other references

Code clause 8.54G(3)(b) and (c)

WHO: Which asset owners provide information for selection

Clauses 2, 3

8.54G(3)(d) WHAT: What information asset owners must provide

Clause 4;

Schedule 1 (information)

Data specification

8.54G(3)(e) WHEN: When the information must be provided

Clause 5;

Schedule 1 (information for selection process)

Schedule 2 (terms and conditions for operations)

Extended reserve manager notification of data request

8.54G(2)

8.54G(3)(a)

8.54G(3)(f)

8.54(H)

HOW: How extended reserve is selected (the basis for selection)

Clauses 6 – 9;

Schedule 3 (default values)

Schedule 4 (model formulation)

Schedule 5 (procurement schedule template)

8.54G(3)(g)

Operational requirements

TERMS: Default terms and conditions that extended reserve providers must follow

Clause 10;

Schedule 2 (terms and conditions)

Schedule 1 (information for periodic monitoring)

Data specification

ERM Calendar

8.54G(3)(h) PAYMENTS for extended reserve providers

Clause 11;

Schedule 5 (procurement schedule template)

Schedule 6 (payments)

The extended reserve manager also proposes to publish:

1. A data specification providing detailed information requirements (including formatting

requirements and guidance on information sources to be used). The draft data

specification is attached as Appendix D.

2. A Data Request for an Extended Reserve Selection Process, being a notification of the

request for information and indicative timeline of a selection process.

3. The ERM calendar, providing deadline dates for information provision during operations

to inform ongoing monitoring of the selected demand units.

Page 10: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 10 of 91

5 Description of the methodology

5.1 Key features of the methodology

The key features are:

1. Use of demand units as the basis of efficient procurement.

2. Use of historic half-hourly information as a proxy for future performance.

3. Reliance on consistent and accurate datasets.

4. Improved transparency.

5. Payments for extended reserve.

5.1.1 Use of demand units as the basis of efficient procurement

This methodology follows the Authority’s intention when creating the extended reserve Code

amendments. It follows the principle in the Code (8.54H(2)(b)), evaluating cost-effectiveness

against relevant costs to make an efficient procurement. In doing so it moves to efficient

procurement of suitable load across the North Island. The unit of assessment becomes the

demand unit (a controllable unit of load). The resulting provider obligations relate to the demand

unit, not to the AUFLS block.

This is a major departure from the existing AUFLS arrangements, which are provider-centric.

The unit of assessment under the current arrangements is the provider. Each provider has an

obligation to provide a portion of their load on each of two AUFLS blocks.

5.1.2 Use of historic information as a proxy for future performance

The methodology relies on historic information to be a proxy for future performance against the

technical requirements. It also relies on average half-hourly (per trading period) information as a

proxy for instantaneous load. So the methodology cannot guarantee to meet the ‘at all times’

requirement set out in clause 6 of the TRS for future years.

The methodology seeks to meet the TRS requirements by using four years of historic trading

period load data to model the variability between years that is likely to be experienced in future

years.

The methodology relies on the assumption that half-hour averages of past load profiles are a

reasonable indicator of moment-to-moment real-time load. If a selection can perform within the

required AUFLS block thresholds1 for four historic years of trading periods, it is expected to be

able to perform well in future years. Change in both load profiles and in networks occurs over

1 The AUFLS block thresholds are set out in the TRS.

Page 11: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 11 of 91

time and this change is expected to occur at similar rates of change seen in previous years. To

allow for changes, additional stand-by demand units are proposed to be selected. These will top

up AUFLS supply levels if needed.

The extended reserve manager considers that there will be some residual risks in respect to this

methodology and any resulting procurement schedule in terms of ability to meet the ‘at all times’

TRS requirement. For example, concurrent unplanned outages, or wide fluctuations in load

within a half hour may occur. Use of estimated data to cover gaps in the historic information

introduces risk. Also if rolling outages or if the Transpower peak load demand response

mechanism were occurring, the percentage of AUFLS supply on each AUFLS block could move

around significantly. This methodology does not allow for (or place responsibility for) adjustment

of AUFLS in real time.

An enhancement to the currently proposed extended reserve scheme to move toward

addressing the ‘at all times’ requirement would be a real-time collection of trading period

information to make the selection and to monitor and manage the selection.

Real-time monitoring and management would provide real-time visibility into AUFLS

performance and the best possible assurance that the procured demand units meet the

AUFLS block thresholds at all times.

However, this enhancement would require significant investment in SCADA2 networks

by some asset owners, compatible communication protocols to be adopted by all parties,

provision of real-time information from all asset owners to a central party, and the ability

for at least some demand units to be armed or disarmed either by a central party or by

prompt real-time response by extended reserve providers to notification. While a few

asset owners have indicated they have the technical capability to provide real-time

information, others have not. The computational requirements to receive and analyse

data for automated decision making on arming and disarming demand units takes time

therefore the ‘at all times’ requirement would still not be met by this enhancement.

This enhancement would require the extended reserve manager to have a system able

to operate 24/7 with a high degree of availability and support services able to promptly

deal with issues (such as metering faults and model infeasibilities) in a prompt manner.

This would be a very substantial increase in the cost of providing this service.

Resolution of these issues would take longer than proceeding with the proposed

approach, delaying achievement of the recognised benefits3.

2 Supervisory control and data acquisition (SCADA) is a system used for the control and indication of equipment.

3 Refer to the cost-benefit analysis in the Authority’s consultation paper, accessed at http://www.ea.govt.nz/

development/work-programme/risk-management/efficient-procurement-extended-reserve/consultations/#c7618

Page 12: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 12 of 91

5.1.3 A consistent, accurate dataset is essential

The methodology relies on an accurate, consistent, and full dataset from asset owners to work

as intended. The data is crucial in two particular respects:

a. Accurate customer class information provided for each demand unit is essential to

enable the extended reserve manager to achieve an efficient procurement, and

b. Accurate load information is essential to have confidence that the dataset provided

sufficiently matches the load that would be tripped if an AUFLS event had occurred at

that point.

Several factors beyond all parties’ control compromise this data to some degree:

Incomplete datasets – in the trial data collection process, we discovered some asset

owners do not have fully complete historic datasets, or in some cases, demand units

have existed for less than the four complete years. Estimates are relied on to complete

the datasets.

Subtraction of interruptible load (IL) – IL cannot be double-counted for extended reserve,

so must be subtracted from each trading period in the historic data. The most accurate

subtraction method is also the highest-effort method for asset owners. We propose this

method (subtraction of cleared IL profiles) to minimise data quality risk.

5.1.4 Improved transparency

A key benefit of the methodology is it achieves a greater level of transparency in the provision of

AUFLS as extended reserve.

The proposed three-monthly updates of load profile information, while still using historic

information, will provide a better level of information and transparency for the Authority and

system operator in regard to extended reserve provision than is available currently.

5.1.5 Payments for extended reserve

The methodology proposes that extended reserve providers are paid for extended reserve, as

allowed for in the Code. This is another major departure from the current AUFLS arrangements.

Payments are proposed because it spreads the cost of extended reserve provision more evenly

across all North Island consumers. Refer to section 10.

Page 13: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 13 of 91

5.2 Overview of the selection process

5.2.1 Inputs to the selection process

Inputs to the selection process:

1. Standard AUFLS provision cost calculation using:

a. Standard AUFLS provision costs

b. Fast response benefit cost

2. Interruption cost calculation using:

a. Standard values of lost load (VOLLs) provided by the Authority

b. The expected interruption hours of each AUFLS block

3. Demand unit information supplied by all relevant North Island asset owners:

a. Unit configuration information (characteristics of the demand unit), and

b. Unit load information (four years of half-hourly load profiles of each demand unit,

comprising at least 60% of the asset owner’s annual offtake)

4. Half hourly offtake information per North Island GXP provided by the reconciliation

manager for each trading period for the same four years.

5.2.2 List of constraints and parameters

Constraints and parameters applied during the final solve process:

To meet the technical requirements:

1. A hard minimum threshold on each AUFLS block

2. A penalty value to each AUFLS block maximum threshold

3. A hard minimum constraint on the target level for each AUFLS block

4. A GXP constraint (if necessary)

To meet the Code principles and operational requirements:

5. No more than 60% of an asset owner’s annual average offtake may be selected

6. Selection of flexible demand units as follows:

a. At least 10% of selected armed load to be designated flexible

b. An additional approx. 10% of the target level to be on standby (initially unarmed)

c. Flexible armed and flexible standby demand units are only selected from those

demand units offered with the ability to arm/disarm remotely by SCADA

d. Flexible demand units are only selected from the pool of demand units with a size of

less than 2 MW

7. A minimum load buffer (to reduce the risk of any one demand unit on temporary outage

resulting in an AUFLS block violation).

Page 14: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 14 of 91

5.2.3 Selection process description

Iterative sampling and solving process

A complete set of demand unit load and configuration data and other parameter values is sent

to the solver. Any missing years of unit load data are filled in at this stage with an estimate and

added to the dataset.

The analyst sets the following sampling parameters for the sampling and stopping criteria:

1. Number of trading periods in initial sample and minimum sampling distance

2. Number of trading periods the sample is increased by each iteration

3. Optimisation gap for sampling and optimisation gap for the final solve

4. Maximum number of iterations.

The trading periods for the initial sample are selected automatically by taking the trading periods

where the aggregate unit load of all submitted demand units to the aggregate GXP load is the

highest and the lowest. Each trading period must be at least 24 hours away from another.

The model formulation is applied to the initial sample using a bespoke application and a

selection of demand units for each AUFLS block is made by the solver that is within a distance

of the optimisation gap from the optimal solution, which is set high at this stage, around 10%.

Then the ‘procurement schedule’ that results is tested against the full set of trading periods for

the 2 test years. The performance results are very bad – there are many instances of the sum of

the demand unit loads falling outside the AUFLS block thresholds. The trading periods with the

worst violations above or below the AUFLS block thresholds are added to the sample. If more

violations are available than the iteration_size parameter, the minimum sampling distance

constraint is applied. The model formulation is then applied to the larger sample, and so on.

These iterations continue, typically about 10 times, until the sample size remains constant

between two subsequent iterations (because there are no or very few violations of the AUFLS

block thresholds so no new trading periods to add) or the number of iterations in the current

process reaches the maximum number of iterations.

The final sample is then optimised for longer to reach the final optimality gap, of around 1%.

The final procurement schedule will be mathematically within 1% of achieving the optimal (most

cost-effective) grouping of demand units into each AUFLS block for the best technical

performance result over the 2 ‘training’ years. It is possible (depending on the demand unit

dataset provided) that a small number of violations of the AUFLS block thresholds may still be

observed.

The procurement schedule is then exposed to the other 2 ‘test’ years to test the AUFLS block

performance on an unseen dataset. The block performance usually drops slightly on the unseen

years, i.e. the AUFLS block thresholds are violated in a few trading periods.

The objective function in the model formulation aims to minimise the costs of the selected

demand units for each AUFLS block.

Page 15: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 15 of 91

The selection is undertaken across the North Island as a single region.

An iterative sampling approach is quick route to a stable solution compared to pre-selecting a

larger sample size.

The final selection process uses significant computational resource to evaluate millions of

potential solutions. A mixed integer linear program run by a Gurobi solver is used. The final

solve takes a number of hours. Full optimality (where the selected demand units are the least-

cost combination) will almost certainly never be achieved. The optimality gap achieved in

generating the Example Procurement Schedule was <1%.

The extended reserve manager proposes to report to the Authority and system operator as part

of the results the optimality level achieved, the aggregate performance against AUFLS block

requirements, and the sample used as well as providing the procurement schedule.

The result is repeatable providing the same conditions are applied, including use of the same

version of the Gurobi solver, the same number of sample iterations, same datasets, constraints

and parameters. The selection process will be audited against this methodology and the Code

requirements, as required by all software used to provide electricity market services.

The system operator commented in its review of the methodology that the sampling and

stopping criteria should be included in the methodology. The ERM accepts this point and

proposes to modify the presented methodology after consultation to include a mathematical

expression of the sampling steps and stopping criteria as described in early draft form in

Appendix E.

Page 16: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 16 of 91

5.3 Example procurement schedule

We illustrate the impact of the methodology by providing an Example Procurement Schedule

generated with the trial dataset provided by North Island asset owners and is provided in a csv

file in Appendix B of this paper. The example is presented in the proposed procurement

schedule template format set out in Schedule 5 of the methodology.

5.3.1 Example procurement schedule analysis

Appendix C contains a brief analysis of the Example Procurement Schedule results. It includes

an indication of the net cost per annum each asset owner would be liable to receive or pay.

Note: the allocation of costs to beneficiaries is determined by clause 8.67 of the Code. The cost

allocation is not determined by the methodology so is out of scope of this consultation process.

5.3.2 Warning – explanatory note

The selection outcome is significantly affected by:

1. the specific set of distributor data used

2. other data inputs (such as relay costs)

3. the selection methodology application of constraints, and

4. the running of the selection process (type of solver, sampling approach, etc).

Items 2, 3, and 4 will be set once the selection methodology is approved and the selection tool

is audited. However, it is vital to note that the selection outcome is significantly affected

by the specific set of demand units that asset owners choose to submit into the selection

process. To the extent that asset owners provide different demand units into the ‘real’ selection

process, the methodology will produce different results. This is further illustrated in the quantity

requirement discussion in section 6.4.1.

Disclaimer: The Example Procurement Schedule cannot be relied upon to be an accurate

reflection of the outcome of any future extended reserve selection process where the demand

unit dataset is different to the trial dataset submitted, nor where changes to any other

variables occur as a result of consultation or audit.

Page 17: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 17 of 91

5.4 The proposed flexible solve process

During the 5-year operational period (in between selection processes) the extended reserve

manager proposes, as part of regular quarterly monitoring, to run a flexible solve to determine

which of the flexible demand units should be armed or disarmed to maintain the best available

AUFLS block performance (refer clause 11 of the methodology).

The model formulation for a flexible solve is set out in Part 2 of Schedule 4.

The objective is to determine the best set of selected demand units to be armed to meet the

technical requirements (AUFLS block target and block thresholds), while minimising the number

of demand units that have to change arming status.

The flexible solve process feeds all selected demand units into the Gurobi solver along with two

years of recent historical load profile information.

The AUFLS provision costs and interruption costs are removed as they are ‘sunk’ costs at this

operational stage. To allow the solver to use standby units to top up blocks we also remove the

constraint requiring > 10% standby load to be allocated, and remove the > 10% armed flexible

load constraint. Other constraints remain.

All standard (inflexible) armed demand units are fixed to ‘armed’ status. The solver process

juggles the armed/unarmed status of the flexible demand units to find the best technical

performance for the fewest switches.

The process introduces a new parameter: a ‘switching cost’. This is applied when a demand unit

arming status is switched from armed to disarmed or vice-versa to prevents excessive ‘flip-

flopping’ of the arming status of flexible demand units, which has a real world impact on

extended reserve providers. Without this parameter, the flexible solve process has no way of

placing a limit on the number of armed demand units that could be required to disarm or arm to

achieve only a very small improvement in AUFLS block performance.

The switching cost plays against three performance penalties:

Maximum AUFLS block threshold penalty

Minimum AUFLS block threshold penalty

Target AUFLS block penalty.

Penalty values are applied for the minimum threshold and the target to avoid the risk of an

infeasible result. It may not be possible to achieve the minimum threshold in every trading

period and the target using the small group of standby demand units available for each AUFLS

Page 18: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 18 of 91

block. The best result achievable with the available demand units will be seen with the flexible

solve. If too many violations are observed, then more demand units may need to be procured

through a top-up solve (limited selection process).

The penalty cost is applied per percentage point violated in any trading period in all three

penalties. In this way the magnitude of an AUFLS block threshold violation is taken into

account. The solver allows violations if they carry a cheaper penalty cost than the cost of arming

or disarming another demand unit.

The ‘costs’ do not represent real-world dollars. The ratio between the switching cost and the

penalties is the important determinant. For example if the Switching Cost to Penalty % point =

1:5 ratio, with all penalty costs initially set to the same value, the solver would switch the armed

status of 5 flexible demand units before it allows a violation of 1% in any trading period OR a

violation of 0.2% in 5 trading periods (or a violation of 0.05% in 20 trading periods and so on).

The actual penalty costs applied are proposed to be reported along with the results of each

flexible solve.

Variations:

Two variations are under consideration:

1. Varying the ratio between switching cost and penalty costs.

a. The ratio should achieve the most acceptable balance between achieving the

best possible technical performance from the selected set of demand units, and

the number of demand units armed or disarmed each quarter. As the load profile

data is historic and is an average of non-IL load within a trading period the

information is an imperfect proxy for performance in any future moment. As such,

the ‘cost’ of achieving a perfect technical performance can be compared to the

‘cost’ effort involving in extended reserve providers having to arm or disarm

demand units.

b. The value of the three penalties can be varied to reflect the relative importance of

avoiding each type of violation.

2. Potentially a secondary objective could be to bias the solver to arm demand units with

lower interruption costs. The model can include a value that represents the relative

interruption cost of each flexible demand unit. If an interruption cost is included the

solver would bias toward arming the demand units with the lower interruption costs and

disarming those with higher interruption costs.

a. While on the face of it this appears to be more efficient, introducing this

parameter may have negative impacts such as increasing switching rates or

lowering performance rates to get the overall lowest ‘cost’ outcome. With no

interruption cost included the solver has no concept of expensive and cheap

units. That means that all of the switches are done purely to meet the technical

requirements.

Page 19: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 19 of 91

5.5 The proposed methodology for a limited selection process

During the 5-year operational period, the Authority could require the extended reserve manager

to undertake a ‘limited selection process’ under clause 8.54J(3) of the Code.

This could occur if, for example, insufficient AUFLS supply remained on any AUFLS block for

any reason. If required to do a limited selection solve to add to the demand units already

selected, the extended reserve manager has modified the full selection process slightly.

The model formulation for the limited selection process (also called a ‘top-up solve’) is set out in

Part 3 of Schedule 4 of the methodology.

The essential difference is the already selected demand units from the previous full selection

process are able to be ‘fixed’. Additional demand units that best fit any ‘holes’ in the AUFLS

blocks (for the best cost) are then selected to top up the blocks.

The process would involve a data request from asset owners, as set out in Schedule 1, Part 1,

clause 4 of the methodology. The exact data requirements would be advised if this process is

ever triggered. No more detail is provided in the methodology as the requirements would vary

depending on the scope of the Authority’s request. The Code clause 8.54J(4) requires that the

Authority give reasons for initiating this process.

Page 20: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 20 of 91

6 Consultation on aspects of the methodology

In this section we explain key aspects and seek your feedback.

6.1 AUFLS provision costs

The methodology applies a standard set of AUFLS provision costs to the information submitted

for each demand unit to calculate an annual cost of providing the selected demand units. The

annual cost is used in the selection process and is proposed to be used to pay each extended

reserve provider.

6.1.1 The cost calculations

The AUFLS provision costs are listed in Part 3 of Schedule 3 of the methodology. These costs

are used in Schedule 6 where they are grouped into four cost types:

1. AUFLS System Administration Payment, containing the AUFLS system administration

cost, applied once per extended reserve provider and attached to the first demand unit

selected from that provider.

2. Relay Operating Payment, containing the relay test maintenance cost, applied to every

selected demand unit.

3. Relay Capital Payment, containing where applicable any of the capital costs such as the

configuring cost, base relay installation cost, initial relay testing cost, and additional df/dt

cost.

4. Demand Unit Interruption Payment, containing the interruption cost for customers on that

demand unit. This cost is applied to all demand units including flexible demand units in

the selection process.

Proposed procurement decision: All new relays must have df/dt functionality

The methodology requires that any asset owner who submits a demand unit with a new relay, to

install the new relay with both u/f and df/dt functionality.

The selection process assumes this functionality is installed on all new relays. It considers all

new relays as potential candidates for AUFLS block 4 (for which df/dt functionality is a technical

requirement) without application of the additional df/dt functionality cost.

Of note: the base relay installation cost proposed is to cover the purchase and installation of a

new relay including the df/dt functionality, and configuring to the required AUFLS block settings.

Page 21: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 21 of 91

Table 6-1: Application of relay capital costs to demand units

Selection

process

Example situations for demand

units

Configuring

cost (RC3)

Initial

testing

cost

(RC4)

Base

relay

cost

(RC5)

Adding

df/dt

cost

(RC6)

Initial selection

process

Existing relay but is not currently in

extended reserve scheme Y Y N

Depends

(1)

In a subsequent

selection process

Existing relay allocated to an

AUFLS block, reselected for the

same block

N N N N

In a subsequent

selection process

Existing relay on allocated to an

AUFLS block, selected for a

different AUFLS block and has not

already configured and tested that

block setting

Y Y N Depends

(1)

In a subsequent

selection process

Existing relay on allocated to an

AUFLS block, selected for a

different AUFLS block and has

already configured and tested that

block setting

Y (2) N N Depends

(1)

Initial selection

process

New relay and the demand unit is

not currently on extended reserve

scheme

N Y Y N

In a subsequent

selection process

New relay on demand unit

(replacing an existing relay),

demand unit is currently allocated to

an AUFLS block and is reselected

for the same block.

N Y Y N

In a subsequent

selection process

New relay on demand unit

(replacing an existing relay),

demand unit is currently allocated to

an AUFLS block and is selected for

a different block.

N Y Y N

(1) The additional df/dt cost is only applied to existing demand units if selected for AUFLS block 4 AND

the relay does not already have df/dt functionality.

(2) The configuring cost is applied to an existing relay as it is assumed that there is some actual cost

incurred by the extended reserve provider as identified by Beca excluding testing costs for switching

a relay from one AUFLS block setting to another.

Page 22: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 22 of 91

6.1.2 Use of standard values is proposed

The extended reserve manager proposes to use standard values representing the average cost

of various capital and operating costs, including administration and transaction costs.

Rationale: Using average expected industry costs is more efficient than the alternative. The

average costs in the report prepared by Beca (see next section), despite the noted variability in

actual costs experienced by asset owners, adequately represent industry costs.

Alternative: The alternative is to request each asset owner to calculate and provide a unique

provision cost for each submitted demand unit into the selection process. We consider this

requirement would place a substantial additional burden on asset owners and due to the early

stage in which the information is required, asset owners would not necessarily be in a position

to determine costs accurately. The extended reserve manager cannot efficiently review the

accuracy of unique submitted costs and this option risks gaming by some asset owners and

risks driving up the overall cost of the scheme.

6.1.3 The proposed set of standard costs

Following feedback from asset owners at workshop 2 that the initial set of costs was too low,

Beca was engaged to survey a sample of North Island asset owners to estimate realistic

generic costs for AUFLS provision. All asset owners were given the opportunity to take part, and

some were specifically approached to obtain costs from a range of asset owners. The survey

occurred in February 2016. Beca’s report ‘North Island Extended Reserve – Automatic

Under Frequency Load Shedding Scheme Item Costs’ dated 7 April 2016 was circulated and

discussed at workshop 3, and is available on the extended reserve manager project webpage4.

Beca’s report provided a recommended generic industry value for each identified cost category

and discussed reasons for variability between providers.

The extended reserve manager proposes to use most of the reviewed costs. The extended

reserve manager’s adoption and treatment of the costs is set out in Table 6-2.

We seek your feedback on both the proposed and omitted costs in this consultation process,

and on whether this set of costs covers the expected costs of AUFLS provision, particularly

noting the technical requirements. The feedback from workshop 3 suggested a number

supported the generic costs, though a range of views are held.

4 http://www.nzxgroup.com/who-we-are/business-overview/nzx-energy/consultations-submissions/extended-reserve

Page 23: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 23 of 91

Table 6-2: AUFLS provision costs used in the selection process5

Values supplied in the Beca report Extended reserve manager use of the values in the selection process

Item

No. Item

Opex/

Capex

Recommended

Medium Cost

Annualised

value Rationale

1a Relay flexible

service cost Opex $220 Not used

Beca’s cost was based on one arming and one disarming of a demand unit per year.

The cost reflected the average of arming and disarming demand units by a technician

sent to site, as well as by remote control.

The extended reserve manager now proposes to select flexible demand units only

from those with remote arming/disarming capability.

Most asset owners surveyed who had remote SCADA control over their demand units

estimated the arming cost at $0.The cost for arming or disarming a demand unit by

remote control is considered insignificant and is removed to reduce complexity.

1b Relay flexible

conversion cost Capex $3,520 Not used

Beca calculated the cost of changing the armed status of a relay within 7 days.

This cost category is not required as it is proposed to select demand units for flexible

service only from those that are flexible-capable already.

2 Relay reconfiguring

cost Capex $6,470 $1,533.39/yr This is the annualised value of Beca’s recommended cost.

3a Relay testing cost -

initial Capex $12,370 $2,931.69/yr This is the annualised value of Beca’s recommended cost.

5 Capital costs are converted to an annual cost over 5 years at a pre-tax nominal rate of 6%, which corresponds with the rates used by the Commerce Commission for

electricity distribution businesses (EDBs). The formula is set out in Part 3 of Schedule 3 in the methodology.

Page 24: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 24 of 91

Values supplied in the Beca report Extended reserve manager use of the values in the selection process

3b Relay testing cost -

maintenance Opex $1,130 $99.50

This is the net present value of Beca’s recommended cost of $1,130 in 10 years’ time,

assuming a nominal interest rate of 6% and an average rate of inflation of 1.5%.

The cost is intended to cover the AUFLS component of a future cost of retesting a

demand unit every 10 years.

4 Base relay cost Capex $21,660 $5,133.42/yr This is the annualised value of Beca’s recommended cost.

5 Additional df/dt relay

cost Capex $17,150 $4,064.55/yr

This is the annualised value (see note 1) of Beca’s recommended cost for adding

df/dt functionality to an existing relay.

It is applied to an existing relay on any demand unit that is identified as not having

df/dt capability, only when the demand unit is considered for AUFLS block 4.

6 Fast response

conversion cost Capex $27,000 Not used

This is the cost to enable a demand unit capable of tripping within a standard

response rate to be converted to respond within 250ms.

To achieve a faster response rate, Beca confirms that generally a circuit breaker must

be replaced.

The extended reserve manager does not consider it efficient to require an asset

owner to replace a circuit breaker so this cost is not proposed to be applied.

7 Implementation cost Capex $19,850 per

distributor Not used

The extended reserve manager raised this as a possible cost category and asked

Beca to investigate it. The cost is not proposed to be used in the selection process.

The rationale is provided below this table.

8a

Relay admin cost:

- Data setup and

entry

Opex /

yr

$3,360 per

distributor $10,080/yr

Beca investigated several potential elements within the administration cost category.

This is the only element that is proposed to be used.

Beca’s cost assumed a yearly collection of load profile information.

As the operational design proposed requires data collection each quarter, the

extended reserve manager assumes data collection costs will increase, but not by as

much as four times as some efficiencies will be gained through a more regular data

provision process. The extended reserve manager multiplied the value in Beca’s

report by 3 to arrive at this cost.

Page 25: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 25 of 91

Values supplied in the Beca report Extended reserve manager use of the values in the selection process

8b

- Incremental

general annual

reporting

requirements

Opex /

yr

$3,030 per

distributor Not used

All industry participants have annual reporting requirements, which are regarded as a

business cost of participation in the electricity industry. The extended reserve

manager, after consultation with Authority staff, considers that an AUFLS component

of general reporting should not be considered a valid AUFLS provision cost.

8c

-Yearly costs for

monthly processing

of AUFLS provision

payments by

clearing manager

Opex /

yr

$4,700 per

distributor Not used

Dealings with the clearing manager including receipt of payments, invoices, and

prudential arrangements (if necessary) are regarded as a business cost of

participation in the electricity industry. The extended reserve manager, after

consultation with Authority staff, considers that dealings with the clearing manager

should not be considered as a valid AUFLS cost.

8d - outage notification Opex $2,460/notice Not used Not used as outage notifications are not required.

Page 26: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 26 of 91

6.1.4 An ‘implementation cost’ is not applied

An implementation cost was not presented in the indicative selection methodology published

with the Authority’s extended reserve Code consultation pack in 2014. It was mooted in the

second industry workshop in 2015 and Beca surveyed asset owners on this cost.

The extended reserve manager proposes not to apply an implementation cost to the selection

process or payment schedule.

Description: Extended reserve providers are likely to incur some one-off costs in setting up to

provide AUFLS under the new extended reserve scheme. Beca’s report identified an average

$19,850 per provider. The costs could be for management time, network design, data analysis,

interest during implementation, set-up of operating procedures and service legal agreements

with the grid owner if the grid owner provides the AUFLS relay. Note this can only include time

once selected, not pre-selection.

If this set-up cost is set to each extended reserve provider as they enter the scheme, in the

initial selection process the impact would be neutral to the selection process as all providers are

in an equal position. The main impact would be on beneficiaries if they are paying for the

additional cost in the first year of operation.

However, in a future selection process, any asset owner not selected the first time around would

be more expensive to select than those already in the scheme. The selection process would be

biased against selecting from ‘new’ or ‘late’ entrant asset owners. The per-distributor AUFLS

administration cost also introduces a limited bias away from selection of smaller asset owners.

Rationale: We consider that application of this cost to the selection process causes unintended

consequences that are best avoided. The future selection bias against new or late entrants is a

sufficiently high barrier to create an intergenerational equity issue in the selection process.

Alternatives:

1. The cost could be annualised over five years instead of fully applied (and paid) in the

first year. However, as a one-off operational cost this is not considered to be

appropriate.

2. The cost could be reduced to be less of a barrier to entry, but to what amount and under

what rationale is not obvious.

3. The cost could be ignored in the selection process, but applied to the payment process

(if adopted). Late entrants would not suffer the bias, but would get paid. However it is not

clear that this approach would satisfy the Code requirement for the extended reserve

manager to procure cost-effectively.

Page 27: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 27 of 91

6.1.5 The ‘relay fast response incentive’ is not applied

Description: A fast response incentive payment featured in earlier versions of the selection

methodology, applied to (paid to) fast-responding demand units capable of meeting a 250

millisecond tripping speed.

Beca did not investigate this cost as it does not relate to any actual cost incurred by a provider,

rather it is an incentive payment and the size of the incentive payment is arbitrary.

The earlier value used was $100 per relay. This is very small compared to the average cost of

converting a AUFLS system from a standard tripping response to fast response of $27,000 per

demand unit to cover the cost of replacing a slower tripping circuit breaker with a faster tripping

circuit breaker.

As originally proposed, this annual operating cost would be paid to extended reserve providers

who already have fast-responding relays as a ‘sweetener’ in recognition of the enhanced

service they provide. It was intended to be an incentive to encourage more fast response

demand units.

Rationale: The rationale for not including this incentive payment is that the disparity between the

cost of fast response conversion and the ‘incentive’ payment does not provide an incentive to

convert demand units. We are interested in asset owners’ views on whether this payment, at

any level, would be considered an incentive.

Alternatives: Reintroduce the cost and/or alter the size of the incentive. Feedback is welcome.

6.1.6 The relay flexible service cost is not applied – is it an incentive?

As noted in the above table, as a material cost this category is proposed to be dropped as we

propose to select flexible demand units only from those with SCADA remote control, and the

cost of service is expected to be negligible as a result.

As our proposal was not raised in workshops, we are particularly interested in your feedback

dropping this cost category, or retaining it and if it is retained as a cost, at what level. In

addition, we welcome feedback on whether retaining the cost would provide an incentive to

encourage asset owners to install SCADA remote control on relays.

The trial data shows that 334 demand units submitted had ‘SCADA indication and control’ but

did not have flexible capability to respond remotely by SCADA (26 of these demand units had

panel control, 134 had manual control, 173 had no flexibile capability). As an incentive payment

payable to relays submitted with SCADA remote control, the extended reserve manager could

set the cost at approximately $100-$200 per relay. The overall impact on the selection cost

would be to increase it by the adopted cost multiplied by the number of relays selected for

flexible service. The number of selected flexible demand units would drop back to around 10%

for flexible armed and for flexible standby.

Page 28: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 28 of 91

6.1.7 The costs are intended to cover technical requirements costs

The costs of AUFLS provision presented in the methodology are intended to cover the costs to

extended reserve providers of meeting the relevant technical requirements in the system

operator’s draft TRS.

At the time of Beca’s survey asset owners had viewed the system operator’s proposed testing

requirements in the extended reserve workshop 2 in November 2015. However changes may

have occurred since then. For example, any costs associated with event capture functionality

are not applied.

We invite comments on any missing or inaccurate costs (including dollar values).

Q1. Do you support the adoption of standard values for AUFLS provision costs in the selection

process?

Q2. Do you support the proposed set of cost categories? Please comment if you consider any

cost categories that are not included should be included and why, and vice-versa. Specific

comments on any cost categories are welcome.

Q3: Do you support the proposed cost values? Specific comments on any values are welcome.

Page 29: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 29 of 91

6.2 Interruption costs and their application

The methodology calculates an interruption cost for each demand unit, to represent the value of

lost load (Voll) should that load be tripped in an AUFLS event.

6.2.1 The interruption cost calculation

The interruption cost is calculated using the customer class proportions submitted for each

demand unit, multiplied by the average annual quantity (MW) on that demand unit, multiplied by

the expected interruption hours (EIH) for each AUFLS block. The cost drops significantly

between blocks 1 and 4 due to the EIH.

Table 6-3: Example of interruption cost calculation

MW size Customer class

Block 1

(EIH 0.278)

Block 2

(EIH 0.136)

Block 3

(EIH 0.054)

Block 4

(EIH 0.032)

1 MW 100% residential

($15,900 * 1MW) * 0.278

= $4,420.20

($15,900 * 1MW) * 0.136

= $2,162.40

($15,900 * 1MW) * 0.054

= $858.60

($15,900 * 1MW) * 0.032

= $508.80

1 MW 50% public health & safety

+ 50% commercial

(($100,000 * 0.5MW) + ($37,200 * 0.5 MW))

* 0.278

= $19,070.80

(($100,000 * 0.5MW) + ($37,200 * 0.5 MW))

* 0.136

= $9,329.60

(($100,000 * 0.5MW) + ($37,200

* 0.5 MW))

* 0.054

= $3,704.40

(($100,000 * 0.5MW) +

($37,200 * 0.5 MW))

* 0.032

= $2,195.20

Q4: Do you support the extended reserve manager’s proposed method for calculating the

interruption cost?

Page 30: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 30 of 91

6.2.2 The interruption costs

The generic values used as expected interruption costs of a 2-hour unanticipated outage by

industry group were determined an industry survey carried out in 2012 by the Authority. The

Authority engaged PricewaterhouseCoopers (PwC) to analyse the results to determine a

generic Voll for bands of customer classes. The 2015 PwC report is available on the Authority

website6. The extended reserve manager received the grouped customer class categories with

the values rounded.

The extended reserve manager has no discretion to alter the industry values determined by

survey. Any comment on them will be regarded as out of scope of the methodology but will be

reported back to the Authority.

The PwC report does not include a Public Health & Safety customer class. The Authority

provided a figure of $100,000 that is sufficiently higher than that of other categories to bias

selection away from other classes due to its sensitivity. As this value was not set by the survey,

the value is discussed in the next section.

6.2.3 Expected interruption hours (EIH)

The Authority published a short paper, ‘Approach to determining AUFLS block return periods’

dated 5 November 2015, in the extended reserve workshop 2 handout pack7.

This paper sets out how the Authority determined the expected interruption hours (EIH) for each

AUFLS block. The extended reserve manager has no discretion to alter this paper. We will

forward any feedback to the Authority.

6 Refer to: http://www.ea.govt.nz/development/work-programme/risk-management/efficient-procurement-extended-

reserve/implementation/report-on-value-of-lost-load/

7 Refer to: http://www.ea.govt.nz/development/work-programme/risk-management/efficient-procurement-extended-

reserve/events/papers-from-extended-reserve-workshop-2/

Page 31: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 31 of 91

6.3 Can the selection process be based more on customer class?

Participants asked us at workshop 3 whether the selection process could be more heavily

based on the customer classes, e.g. by reducing or removing the relay costs as they appeared

to influence the selection more than the interruption costs.

We ran some scenarios to alter the balance of relay and customer interruption costs. Table 6-4

shows the overall impact on the selection result. The scenarios are all comparable to the

indicative procurement schedule presented at workshop 3.

The results show that reducing the relative size of the relay costs results in a selection result

that more clearly favours the lower cost customer classes. The public health and safety (most

expensive) class is shown in the table as an indication. When relay costs are dropped entirely,

the selection is manifestly less cost-effective, as the selection process selects a much larger

number of demand units.

Table 6-4: The impact of changing the costs on the selection result

Scenario Selection

cost

Proportion of

interruption cost

to relay cost

# demand

units

selected

% public

health &

safety

selected

# selected

DUs with

100% PH&S

Workshop 3:

Standard relay costs

Standard interruption costs

$6,216,438 IC 44% /

Relay cost 56% 556

5.7% in 277

DUs 9

All relay costs halved

Standard interruption costs $4,690.762

IC 58% /

Relay cost 42% 617

3.3% in 297

DUs 5

All relay costs quartered

Standard interruption costs $3,564,070

IC 71% /

Relay cost 29% 642

1.5% in 303

DUs 0

No relay costs

Standard interruption costs $2,307,157

IC100% /

Relay Cost 0% 987

0.98% in 464

DUs 0

Standard relay costs

Standard interruption costs

with doubled public health

and safety cost ($200,000)

$6,532,171 IC 43% /

Relay cost 57% 588

2.8% in 283

DUs 3

We also examined on what basis we could modify the cost inputs. The methodology must meet

the Code requirement for cost-effective procurement (clause 8.54H(2)(b)), which means that

both cost types (provision cost and interruption cost) must be considered in selection.

Page 32: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 32 of 91

6.3.1 Treatment of public health and safety customers

Description of issues: The selection methodology cannot completely avoid the selection of

public health and safety customers. Many are embedded within demand units containing other

customers (53% of submitted demand units contained some proportion of public health and

safety customers).

A large number of the submitted demand units containing public health and safety customers

are already on the existing AUFLS scheme, suggesting the presence of these customers was

not currently considered critical enough to be avoided.

A relatively high price signal applied to these customer results in their placement in AUFLS

blocks 3 and 4 that are less likely to trip. Of the total load-weighted quantity of public health and

safety load selected in the Example Procurement Schedule, 50% was selected on AUFLS block

4, 30% on block 3, 15% on block 2, and just 5% on block 1.

The extended reserve manager considered the arguments for raising the cost to provide a

stronger price signal for this sensitive group into the selection process, or for lowering it as

many of these customers have made provision for back-up supply, so their true interruption cost

can be regarded as lower. Industry participants raised these issues in workshops.

We also noted that in the trial dataset distribution companies inconsistently allocated customers

into the public health and safety class.

Proposal: In consideration of these issues, the extended reserve manager proposes:

1. To provide more precise guidance on which customers to place in the public health

and safety category, to achieve better consistency across distribution companies. This

guidance is set out in the data specification. A customer must meet 2 criteria to qualify

for this category: one, they must be on PROPS 1 and 2 or in the lifelines definition, and

secondly their ability to carry out their public health and safety activities must be

potentially compromised by disruption to supply in an AUFLS event. A customer known

to have full and tested back-up supply would be unlikely to meet the second criterion.

2. To not allow asset owners to submit a demand unit that contains 100% public health

and safety load (refer Schedule 1, Part 1, clause 3 of the methodology)8.

Rationale/benefit: Excluding demand units with 100% health and safety customers is the

only way to be assured that a demand unit that is solely from this category will not be

selected. Demand units that are 100% public health and safety tend to contain a larger

user, such as a metropolitan airport or port. As these kinds of loads are not suitable for

selection, then they should not be eligible for submission.

8 11 demand units in the trial dataset were in this situation and they were excluded from the Example Procurement

Schedule dataset.

Page 33: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 33 of 91

Cost: is the opportunity cost of not being able to include these demand units in the

selection process. The cost is relatively small, particularly as providers are likely to not

submit some of these demand units due to their sensitivity anyway. Table 6-5 provides

an example of the impact of including these excluded demand units.

Table 6-5: Example cost of excluding 100% public health and safety customers

Scenarios Selection cost

Number of 100%

PH&S demand

units selected

Total number of

demand units

selected

Example procurement schedule $5,964,535 0 572

Scenario including the submitted

11 demand units that are 100%

public health and safety

$5,921,477 4 558

3. To retain the Authority’s proposed interruption cost of $100,000 in recognition that

this group of customers are regarded to be less desirable for AUFLS.

Rationale: In the absence of a quantitative evidence for raising or lowering the proposed

value of lost load of $100,000, the estimate originally proposed by the Authority is

considered appropriate to retain. The relatively high price signal acts to some degree to

reduce selection of public health and safety and it relegates public health and safety

more to blocks 3 and 4, which are expected to be interrupted less often.

Q5. Do you support the extended reserve manager’s proposal to adopt a generic value of

$100,000 for the public health and safety customer class and to tighten the definition of public

health and safety as set out in the data specification?

Q6: Do you support the proposal that demand units with 100% public health and safety

customers on them are not eligible for submission?

Page 34: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 34 of 91

6.4 Demand unit information requirements

6.4.1 Quantity requirement

The minimum unit load quantity requirement is 60% of each asset owner’s annual average

offtake for four historical years.

6.4.1.1 The 4-year requirement

Description: For selection of extended reserve, each asset owner is to provide (in demand units)

at least 60% of their annual average load (calculated after the subtraction of any interruptible

load (IL) from the demand units) for four historic years. The quantity requirement is measured

on the most recent of the years requested (the reference year). Noting:

1. The submitted load data must have IL subtracted as IL cannot be double-counted for

extended reserve. This is a technical requirement set by the system operator.

2. If an asset owner has a significant portion of its load committed to IL such that less

than 60% of its annual average offtake as seen at the relevant grid exit points (GXPs) is

available for extended reserve, then the quantity requirement is to provide the extended

reserve manager with the remaining available load. (Several directly connected

consumers are in this position.)

3. We encourage asset owners to provide more than the minimum 60% where they

have suitable load as the selection process is likely to achieve a more cost-effective

result with more demand units.

4. Four years of data for each demand unit must be provided where available.

However fewer years may be submitted for a small portion of demand units, to enable

suitable demand units with limited history to be considered.

5. Gaps in the dataset within a year must be estimated by the asset owner in

accordance with the estimation process set out in the data specification.

This quantity was requested in the trial data collection process conducted in late 2015.

Rationale:

The selection process requires 4 years of load profile information to obtain confidence that a

robust selection result is made. The extended reserve manager proposes to use the 2 most

recent years to create the solution, and the older 2 years to test the solution. (Refer to the

AUFLS block thresholds section 6.6.3 for an explanation of why 2 years is used for selecting.)

The older 2 years are also critically important. They will also be used by the extended reserve

manager and system operator to review the draft procurement schedule.

From a data quality perspective, the extended reserve manager would prefer to set a 2-year

minimum limit for each demand unit. The selection is made on the most recent 2 years and it is

less than desirable to select using estimated data (see the next section).

Page 35: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 35 of 91

However, workshop feedback from asset owners was that a 2-year limit would reduce their

options to submit suitable demand units as 2 years of data is not always available where

network changes have occurred.

Consequently, the extended reserve manager agrees it is more useful to receive limited

information on some newer suitable demand units to be able to include them in the process,

than to insist on 2 years of information for all demand units.

However it is very important to submit as many years of information as possible for each

demand unit. If for example an asset owner has large gaps in the information for a year (e.g.

several months of missing data) it is preferable that the asset owner submits the year and fills

the gaps with estimates, so that any actual data that does exist can be used.

6.4.1.2 Missing years must be estimated by the extended reserve manager

The selection process only works with a complete 4-year dataset for every demand unit. This is

because the process adds up the half-hourly load for every submitted demand unit when

comparing selection options. If a year is missing:

from one of the 2 most recent years, then the demand unit is very unlikely to be selected

as its performance will be poor across the 2 years

from the 2 earlier ‘test’ years, then the AUFLS block performance results will suffer due

to missing data, which makes performance very difficult to assess.

Some demand units may not have existed in the earlier years. In estimating data for demand

units that did not exist, the extended reserve manager is treating the years as proxies for future

years where the demand unit is expected to exist. The fit is not perfect but for selection

purposes it is preferable than a year of ‘zero’ information or no information.

To get around the problem, the extended reserve manager proposes to fill in any missing years

by estimating unit load data for those years based on the submitted data. The estimated years

will be identified as such to the system operator, as they need to be treated with more caution,

particularly if relied upon to make the selection. (This is why receiving data from asset owners

for the 2 most recent years is particularly important.)

The estimation process used in producing the Example Procurement Schedule was to take the

nearest future year of submitted data, adjust it to line up week days with the estimated year and

then scale this unit load data based on a ratio calculated between the corresponding GXP

offtake in the future year, and the GXP offtake in the estimated year.

We have since trialled an alternative method: to simply copy the ‘source’ unit load data from the

next future year, adjusting it only to line up the weekdays.

Comparing the two estimation methods against the actual metered load profiles supplied for a

large sample of demand units submitted with metered data we found:

Both methods produce a similar profile to the actual data over the year.

Page 36: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 36 of 91

In most cases the average estimated load produced from both methods differed from the

actual data by 6 to 25% (bigger or smaller) but at times the variance was a lot higher.

Where the GXP loads varied between the 2 years, the GXP-adjusted method produced

large variances from the actual data of up to 6 or 7 times.

Consequently, the extended reserve manager proposes to estimate missing years by simply

copying the data from the nearest future year and aligning the weekdays.

We propose to specify this estimation method in the extended reserve functional specification

(which must be approved by the Authority and published on the Authority’s website). The

system operator has suggested that the estimation method be included in the selection

methodology. We would prefer not to include it in the methodology so that we can improve the

estimation method over time without having to re-consult on the methodology.

Q7. Do you accept the provision of 4 years of data as the minimum quantity requirement for

load profile information where it is available?

Q7a. Do you support the proposed method for estimating missing years of data and its inclusion

in the extended reserve manager functional specification?

6.4.1.3 The 60% offtake requirement

60% of North Island annual average offtake provides a sufficient number of demand units to

have a high level of confidence that the selection process will achieve a technically feasible

result. However, if a higher quantity is submitted, the selection process is likely to achieve a

more cost-effective result.

We illustrate the importance of obtaining as large a dataset as possible from asset owners (i.e.

preferably more than the minimum 60% required). We reduced the trial demand unit dataset in

four ways. In the first three scenarios we reduced the dataset to just the bare minimum

requirement of 60% from the asset owner’s average annual offtake, and in the fourth we

reduced it further:

1. Largest demand units removed from asset owners who provided more than 60%

2. Smallest demand unit removed from asset owners who provided more than 60%

3. ‘Best guess’ removal of demand units that asset owners might choose not to provide

4. The ‘best guess’ scenario was reduced further to just 55% of each asset owner’s average

annual offtake.

Table 6-6 shows that the smaller the dataset and the smaller the average size of the demand

units in the dataset, the higher the total selection cost.

Page 37: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 37 of 91

Table 6-6: Selection costs using reduced datasets

Scenario Average size of

selected demand unit

Total annual

selection

cost9

Up to 100% of offtake submitted: Example Procurement

Schedule 1.68 MW $5,964,535

1. 60% submission limit: Largest demand units removed 1.26 MW $7,824,406

2. 60% submission limit: Smallest demand units removed 1.81 MW $6,429,323

3. 60% submission limit: ‘Best guess’ demand units

removed 1.52 MW $6,596,486

4. 55% submission limit: ‘Best guess’ demand units

removed 1.38 MW $7,365,336

The reduced datasets still produced a procurement schedule that meets the technical

requirements. No performance violations or N-1 performance violations were observed from the

2014 or 2013 years. All block 1 & 2 targets were met at 10%, and block 3 & 4 targets were met

with slight variations ranging 6.2 to 6.6%.

We are aware that some asset owners struggled to provide 60% of suitable load from their

networks. Placing a higher requirement may be too onerous for some providers to meet, and/or

may push them into submitting load that is less than suitable for extended reserve (noting we

also propose to not allow demand units with 100% public health and safety customers).

Also note in clause 1(b) of Part 1 of the methodology that we have put in a back-stop in case

the dataset we receive is insufficient to produce a good result. If the extended reserve manager

requires additional information during a selection process, we will issue a request and asset

owners must respond within 20 business days. The intention of this clause is to provide for

flexibility to extend the 60% quantity requirement if necessary.

Benefit of 60% requirement

With up to 40% of load not being required to be submitted, unsuitable load can be withheld from

the selection process. Unsuitable load may be load that the asset owner has already committed

to PROPS10 or other uses, or load used by public utilities or by any particular customer

(regardless of customer type) whom the asset owner considers unsuitable for AUFLS. By

having some discretion to withhold some load, distributors can make these common-sense

decisions to avoid selection of these loads.

9 For 5 years (the capital cost component is paid back after 5 years).

10 Participant rolling outage plans, which may require up to 25% of load.

Page 38: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 38 of 91

A key aspect of this methodology is that asset owners make the ‘first cut’, which is important as

they are the best informed about their network requirements and sensitive customers and the

selection methodology cannot take these nuances into account. The extended reserve manager

expects asset owners who have room for discretion to withhold any load that they consider

would not be suitable as long as the 60% data requirement is met.

Consequently, we propose to hold the requirement to ‘at least 60%’ as this is sufficient to obtain

a reasonably cost-effective result against a similar dataset, and to encourage asset owners to

submit more if they have suitable load.

Limitation of the 60% requirement

Asset owners who already have a significant portion of their load committed to IL have no

discretion to withhold their remaining load.

However in some cases the remaining load may not be suitable for AUFLS. Some is understood

to require multiple relays per demand unit.

The extended reserve manager has not come up with an appropriate ‘exemption’ rule that

would deal with each unique situation. Consequently some asset owners are required to submit

information on potentially unsuitable load. The alternative available to these asset owners is to

consider the relative costs and benefits of making their IL load available for extended reserve.

Alternatives:

Two alternatives are considered and rejected for the following reasons:

Request less than 60% load or less than four years – a smaller dataset will be less cost-

effective as the cost of the solution will increase from a smaller dataset, and the risk of

an infeasible or less robust solve increases.

Request all non-IL load (up to 100%) – this alternative has been discussed at workshops

and potentially may be supported by some participants. The risk in this approach is that

distributors will not be able to exercise some discretion to withhold unsuitable load and

that there may be inefficient cross over for extended reserve provider management of

other power system requirements such as PROPS and manual load shedding.

Important note: A smaller average demand unit size raises selection cost

The smaller the average size of submitted demand units, the higher the overall selection cost

(and the more all beneficiaries are liable to pay).

Scenario 1 in Table 6-6 illustrates the sensitivity to the overall cost of the scheme to the removal

of the larger (cheaper) demand units – the selection cost increased to $7.8 million per annum

(for 5 years).

This is a possible scenario. If asset owners in general submit only the minimum quantity of

demand units and the average size of demand units drops, the extended reserve manager

Page 39: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 39 of 91

estimates it would be possible to obtain an overall selection cost of approximately this amount

applying the methodology for the first selection process.

In the trial process, one asset owner submitted significantly larger demand units than others

(average size of 7.6 MW), and as a consequence in the Example Procurement Schedule is both

selected up to the selection cap of 60% and is a relatively large net payer into the scheme.

We modelled the potential impact if this asset owner changed their demand unit submission to

smaller demand units with new relays. We split their large demand units into 6 identical demand

units with identical load profiles each one-sixth of the original size (average 1.27 MW) and fed

this dataset change into the selection process. This scenario used the full trial dataset. The

results were:

The asset owner’s selected quantity dropped from 60% of its annual average offtake to

54% of its offtake. The quantity selected from other asset owners increased slightly.

The overall average demand unit size of submitted demand units dropped from 1.1 MW

to 1.03 MW and average selected demand unit size dropped from 1.68 MW to 1.43 MW.

The total allocable cost (selection cost) of this scenario increased to $6,464,571 per

annum ($500,000 more expensive than the example procurement schedule).

The asset owner’s indicative net payer allocation position of approximately $150,000 in

the Example Procurement Schedule moved to an indicative net payee position of

approximately $240,000. Almost all other asset owners’ indicative net cost allocations

worsened as they picked up the resulting additional $890,000.

Q8: Do you support the requirement for asset owners to provide at least 60% of offtake (the

60% to be net of interruptible load) in demand units?

Page 40: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 40 of 91

6.4.2 Information description and the data specification

The information required from asset owners is set out in Part 2 of Schedule 1 of the

methodology.

The methodology states that each asset owner must provide information in accordance with the

data specification (Schedule 1, Part 1, clause 2 and Schedule 2, clause 9). This includes

providing it via the transfer mechanism (portal) set up by the extended reserve manager for this

purpose.

Some changes to the data specification have been made since the trial data collection process,

mostly in response to feedback from asset owners following the trial process.

In addition, several of the unit configuration fields are no longer required because they will not

be used. A field requesting the ‘zone substation’ (if any) is added as this will be useful for risk

analysis. The format of the unit load file has been altered to have one trading period per row,

and to accept negative values to represent occasional injection of load from a demand unit. The

file formats are set out in the data specification.

Portal changes are being made to reduce information submission time.

The data specification is provided in this consultation pack as Appendix D. Your feedback is

welcome.

Q9: Does the data specification provide clear and achievable instructions that will promote a

consistent and efficient response from asset owners?

Q10: Do you have any other feedback on the data specification?

Page 41: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 41 of 91

6.4.3 The accuracy requirement

One of the Authority’s proposed Code amendments published in August 2016 was to require

asset owners to submit accurate information. The proposed Code amendment states the data

must be accurate as defined in the extended reserve manager’s data specification.

Accuracy for the selection process is particularly important in two respects:

The accuracy of the load profile information of each demand unit affects the AUFLS

block performance.

The accuracy of the customer class information affects the efficiency of procurement.

This section defines what providing accurate information is proposed to mean in this context.

Refer to the data specification published with this consultation pack.

6.4.3.1 Accuracy definition of the load profile information

In an AUFLS event, the actual quantity of demand unit load on each AUFLS block in the

seconds before the event occurred (the pre-event demand) should be within the required

AUFLS block thresholds.

Several issues potentially impact on the accuracy of the load profile information. The issues are

outlined here and alternatives considered, and a preferred method proposed.

1. Historic, half-hourly averaged load information is accepted.

A risk inherent in this methodology (as noted in section 5.1.2) is that historical, averaged half-

hourly (trading period) load information is relied upon to be a reasonable reflection of the

quantity of load on the group of demand units for any AUFLS block in any future moment in

which an AUFLS event may occur.

To achieve the most accurate information available, the extended reserve manager proposes

that the asset owner is required to provide an average value using the greatest number of

sample periods available within the trading period.

The extended reserve manager cannot identify a cost-effective alternative to using averaged

historical trading period information during the selection process.

2. MW load values are calculated by tested equipment.

There are inherent inaccuracies in the measurement and calculation of the demand unit load

figures. Factors influencing the accuracy of the load figures include the quality of current

transformers, voltage transformers, metering equipment, communication and monitoring

equipment. Those asset owners who measure the demand unit in amperes introduce a

calculation error due to the use of generic network voltage and power factor figures. The

extended reserve manager will accept the MW values presented on the basis that asset owners

have installed and tested the various equipment to meet industry standards.

Page 42: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 42 of 91

3. Subtract the curtailable IL half-hourly profile from relevant demand unit half-

hourly load profile.

Description:

The curtailable IL half-hourly profile is the quantity of IL that is available in each half-hour and

potentially curtailed if the IL is tripped. The extended reserve manager proposes that asset

owners sum the curtailable IL half-hourly profiles of relevant ICPs in each demand unit and

subtracts the curtailable IL half-hourly profile from the relevant half hour load value for the

demand unit. Note that:

The curtailable IL half-hourly profile for the reference year (most recent year) only is

proposed to be used. The IL half-hourly profile for this year is proposed to be subtracted

from each of the 4 years.

Where the maximum contracted quantity for an IL contract is less than 120kW, the

maximum contracted quantity may be used.

Refer to section 4.5 of the data specification for further information on the proposal.

The half-hourly profile of curtailable IL should be obtainable from the ancillary service agent or

from the asset owner’s own records where asset owners offer IL directly. Clause 8.54B(3) of the

Code enables asset owners to request additional information from ancillary service agents

relating to IL contracts.

Rationale:

The extended reserve manager considers that the subtraction of IL from demand units is

potentially a major contributor to data inaccuracy. Therefore we propose the most accurate

method to minimise inaccuracy and the resultant risk of violation of the AUFLS block thresholds

during an AUFLS event. We propose the construction and use of a standard profile based on

the reference year to reduce the quantity of work required, including use of the profile during

operations.

The extended reserve manager considers the curtailable IL half-hourly profile provides the most

accurate measurement of IL that is actually likely to trip in any half hour in advance of AUFLS.

The main alternative in our view is the maximum contracted quantity method as it requires less

asset owner effort, although this method is not very accurate (see description of alternatives

below) and would likely result in over-procurement of AUFLS.

We appreciate this is one of the areas that may consume the most effort during data

preparation. We are interested in your feedback.

Potential size of the inaccuracy

In the trial data round, 9 asset owners reported they had subtracted (mostly using the maximum

contracted quantity method) 195 MW from 913 demand units, out of a total 2041 MW average

annual MW submitted from 1904 demand units. The largest reported quantity came from 1

Page 43: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 43 of 91

directly connected consumer, who reported 71 MW of IL on its demand units. The remaining

124 MW was reported by 8 distribution companies.

A reported 107 MW (including from the directly connected consumer) was subtracted from the

selected demand units listed in the Example Procurement Schedule.

Alternatives:

The extended reserve manager discussed various options for subtracting IL with asset owners

in the first extended reserve workshop in September 2015. Alternatives include:

a. Subtract the maximum contracted quantity stated in the ancillary service agent’s

contract, which the agent is required to provide to the connected asset owner under

clause 8.54B of the Code. This method was used by most asset owners for the trial data

process. Subtracting this quantity from each trading period is a conservative method. Its

major advantage is it is relatively simple to do. It mainly avoids any double-counting

between AUFLS and IL. However, the main disadvantage is subtraction of the same

single quantity from every half hour does not profile what is actually available in any half

hour and for most if not all trading periods this quantity is higher than the amount that

would be tripped. So if asset owners use this method, the extended reserve manager will

potentially over-procure AUFLS as the load data supplied by the asset owner will under-

report the non-IL quantity available. An additional disadvantage to this method is in

periods of low offtake from a demand unit, a non profiled maximum contracted quantity

can be larger than the offtake available on that demand unit, resulting in the load profile

unnaturally appearing to be ‘zero’ in these trading periods.

b. Subtract the half-hourly profile of offered or cleared IL, the quantity of IL offered by the

ancillary service agent into the IR market or cleared by the market. Subtracting the

cleared IL profile from the unit load profile is not considered to be an accurate method

for the extended reserve selection process as the IL is not always disarmed when it is

not cleared in the market. Using offered IL may also underestimate the amount of IL

available due to the practice of those offering IL arming more IL than is offered for

compliance reasons11.

c. Calculate the average MW tripped load from the previous five IL events to determine an

estimated IL proportion. This does not capture time of day differences in IL provision and

may under-estimate IL on the demand unit.

11 Ancillary Service Procurement Plan technical requirement B35.2.2 is for the ancillary service agent to reduce

demand or load disconnection consistent with the dispatched quantity of IL whenever the frequency falls to or below

the trip frequency. The Code is silent as to whether or not IL not dispatched should be disarmed. It could be argued

that B35.2.2 implies that IL not dispatched should be disarmed with the words ‘consistent with the dispatched

quantity’ but then this has the proviso that this is only when the ‘frequency falls to or below the trip frequency’ not

when frequency is above the trip frequency i.e. during normal system operation). The extended reserve manager

has been informed that the practice of disarming when not dispatched does not always occur and therefore cannot

rely on the accuracy of the offered or cleared IL.

Page 44: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 44 of 91

d. Use average annual or monthly offered or cleared IL quantities. These methods also do

not capture time of day differences in IL provision and risk double-counting of IL and

AUFLS.

e. Exclude any demand units containing IL from extended reserve. This is not practical,

because it would too greatly limit the number of demand units available for extended

reserve.

Other notes:

Under-frequency events in which IL was tripped occurred on average (2011-2015) 5 times a

year, and the historical dataset will already reflect those periods so the extended reserve

manager does not compensate for these periods when selecting extended reserve.

During operations (after selection) extended reserve providers must continue to subtract IL from

their unit load information for each selected demand unit. The extended reserve manager

proposes that asset owners continue to use the IL profile obtained for the reference year and

subtract the IL profile of the relevant three months from the quarterly dataset, or update the IL

profile where the extended reserve provider is aware the IL quantities have changed.

Q11: Do you support the proposal to require interruptible load to be subtracted using the

curtailable IL half-hourly profile on each demand unit?

4. Estimations in the load profile due to missing data points within a year must be

filled in using the estimation method proposed in the data specification.

Where some of the load profile information is not available within a year, the asset owner must

follow the estimation process set out in the data specification to fill in the missing trading periods

in a consistent manner that is likely to mimic what the actual profile may have been. Providing

this estimation method is followed, the load profile data will be regarded as sufficiently accurate.

Even when a demand unit is missing fairly large blocks of information (several months in a year)

from the extended reserve manager’s perspective it is worth obtaining the unique information for

the months that do exist.

If an entire year of information is not available or if the demand unit did not exist, the asset

owner must not submit for that year. Entire years of zero data are not acceptable.

Our intention is the estimation method used by all asset owners is consistent, achievable, and

accurate.

Providing the estimation method is followed and the load marked as estimated, the data will be

regarded as sufficiently accurate.

Page 45: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 45 of 91

6. Any potential impact of embedded generation within a demand unit is not

considered in procurement.

The impact of embedded generation present in a demand unit is embedded within the load

profile seen at the GXP/demand unit. The risk is whether particular embedded generation trips

off early during an AUFLS event. As the TRS does not require the extended reserve manager to

take account of embedded generation in procurement, no adjustments to the load profile are

required.

7. Data is required to be accurate by the revision deadline.

The extended reserve manager proposes that the accuracy requirement is breached if any data

issues are not corrected by the end of the revision period.

The proposed Code amendment for accurate data12 could imply that a revision process is not

required. However, based on the trial data provision experience, we consider the revision period

is important as the selection process only takes place once every five years. While some data

issues can be picked up by system validations as the files are uploaded, other issues will only

be spotted through ERM analyst review, such as spikes or abnormal quantities of zeroes in the

unit load information.

The methodology proposes that asset owners are required to provide the demand unit

information by the deadlines set by the extended reserve manager.

Workshop 3 feedback from asset owners was that most considered they could provide the data

in 2 months, and revisions in 2-3 weeks. Most people wanted at least 3 months’ notice. We

have attempted to provide several months’ notice by advising of the indicative data collection

dates mid-year.

In addition, to provide asset owners with some assurance of the timeframes, the minimum data

timeframes that the extended reserve manager can set for the selection process are proposed

in the methodology (refer clause 7). They are for data provision at least 40 business days after

the notification, and for data revision (should any revisions or explanations be needed) at least

10 business days.

The extended reserve manager proposes that the accuracy requirement would only be

breached if any data issues were not corrected by the end of the revision period.

Q12: Do you support the data provision timeframes proposed for asset owners during the

selection process of 40 business days for data provision and 10 business days for revision?

12 See issue 3 in the Authority’s 23 August 2016 consultation paper, accessed at

http://www.ea.govt.nz/development/work-programme/risk-management/efficient-procurement-extended-

reserve/consultations/#c16167

Page 46: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 46 of 91

6.4.3.2 Accuracy definition of the customer type information

The customer class data provided in the trial process was supplied on a ‘reasonable

endeavours’ basis as often consumption information was not readily available. As a result, the

information supplied was not consistent across asset owners, with some estimating using

general knowledge of their networks.

The Authority has proposed a Code amendment to enable distribution companies to access

EIEP1 and EIEP3 information produced by traders for extended reserve use. EIEP1 and 3

provides consumption volume information at an ICP level, either at half-hours or non-half-

hourly. This information can be aggregated for the reference year and then the ICPs can be

cross-referenced to the ANZSIC codes to calculate customer type portions on each demand

unit.

The extended reserve manager sets out the proposed method in full in the data specification.

We propose that all distribution companies must use this method and these sources (or

equivalent sources that meet this accuracy standard) to calculate the portion of load consumed

in the reference year by each customer type present on the demand unit.

In addition, we propose a tighter definition of the public health and safety customer type

(discussed below).

Rationale: We consider that all distribution companies will be able to apply this method to

accurately and consistently calculate the load-weighted proportion of each customer class on

each demand unit, providing a good level of accuracy.

Q13: Do the proposed methods for estimating missing data and for customer class allocation

promote a reasonable and attainable standard of accuracy?

Page 47: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 47 of 91

6.4.4 Limitations on asset owners involved

Proposal: The extended reserve manager proposes to not require very small asset owners who

have a total annual average offtake of less than 1 MW to provide information.

Background: The methodology requires information only from North Island asset owners, as

required by the TRS. The relevant asset owners are distribution companies and directly

connected consumers. This currently comprises 15 distribution companies and 7 directly

connected consumers, whose annual average offtake ranges from c. 5MW to c. 980MW each.

It is important to draw the distinction between larger asset owners who submit very small

demand units and a very small asset owner. It is very useful to the selection process to receive

all sizes of demand units including extremely small demand units. Some tiny demand units are

selected in the example procurement schedule. However, the likelihood of selection drops as

the size of the demand unit drops. The likelihood of selection drops even further for very small

asset owners as the per-asset owner AUFLS administration payment is attached to that first

demand unit, causing the relative cost of the first selected demand unit from an asset owner to

be high.

The smallest selected demand unit in the example procurement schedule contains 0.011MW

average annual net load. Out of 445 submitted demand units under 0.5MW, 8 were selected

with a probability of selection of <2%. However, these demand units will be making a positive

contribution to the overall selection performance result and are welcome.

Grid-connected generators also draw down very small loads directly from the grid. A proposed

Code amendment clarifies that consumption from grid-connected generators is not intended to

be considered for extended reserve.

Rationale: Authority staff advise it is practically inefficient for the extended reserve manager to

deal with very small asset owners. Currently one asset owner is in this category, with an annual

average load of c. 0.05MW. A 1 MW limit is considered practical. While only one asset owner

currently falls (well) below this limit, the proposed limit is 1 MW to capture any potential future

small asset owners up to this size.

Benefit: The practical efficiency of saved service provider cost of dealing with very small asset

owners.

Cost: The cost is the lost opportunity from having the additional information in the selection

process. (The impact on other asset owners is considered to be negligible given the very small

likelihood of selection.)

Q14. Do you support the proposal to remove the information requirement for asset owners

whose offtake is less than 1 MW on average per year?

Page 48: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 48 of 91

6.5 North Island offtake information (reference dataset)

The AUFLS block percentages must be measured against a reference dataset. The dataset

proposed to be used is the offtake recorded by the reconciliation manager for each trading

period for each North Island GXP in the relevant historical period. A proposed Code amendment

requires the reconciliation manager to provide this dataset to the extended reserve manager.

The extended reserve manager will also refer to this dataset for the most recent historical year

to determine the average annual offtake of each asset owner and thereby the MW size of the

60% data quantity requirement.

6.6 Constraints applied to meet technical requirements

In this section we describe the constraints and parameters that are applied to the selection

process to seek to meet the technical requirements.

The methodology must specify how the extended reserve manager will procure extended

reserve according to the TRS (refer clause 8.54G of the Code). The relevant technical

requirements are in TRS:

Clause 5 (use AUFLS)

Clause 6 (AUFLS systems must at all times together be capable of providing AUFLS in

the North Island in the AUFLS blocks)

Clause 8 (the AUFLS blocks)

Clause 9 (IL is excluded), and

Clause 16 (prefer fast-response AUFLS systems and take into account the system

operator’s quality levels).

6.6.1 ‘At all times’ requirement

The TRS clause 6 ‘at all times’ requirement is not possible to directly guarantee using this

methodology. As stated in section 5.2.1 this methodology proposes to use historical load

profiles and trading period averaged profiles, to determine a suitable mix of demand units that

meet the AUFLS block requirements in the studied historical years.

6.6.2 Use of penalty costs is minimised

Use of penalty costs in the solve process is kept to a minimum. Penalty costs can discourage

violations of technical requirements as the solve process seeks to minimise cost. If a penalty

cost is added each time a selected group of demand units violates an AUFLS threshold, the

solve process will attempt to find a cheaper/better solution to avoid that penalty.

Page 49: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 49 of 91

If the solver has options to find a middle ground between penalties and ‘real world’ costs it will,

which means an average performance is achieved on all fronts. By minimising use of penalty

costs, there are fewer factors for the solver to try and balance. The result is a better technical

performance for lower ‘real-world’ costs.

In other words, use of penalty ‘costs’ reduce the effectiveness of the ‘real’ costs. Since all

penalties and costs are grouped and optimised together in a single dimension, the more cost

penalties used the fuzzier the results will get. Introducing any penalty value which would be hit

often (such as AUFLS block thresholds) would result in:

real costs having less impact (e.g many more demand units with high amounts of public

health and safety customer load being selected)

AUFLS block thresholds being hit more often (as it can trade off one against the other)

worse technical performance (as the 1% optimality gap is within 1% of the optimal cost

which includes all penalty costs).

Use of penalty costs also slows down the solve process.

Penalty values are used when we have no other option, as in the case of the maximum AUFLS

block threshold penalty.

6.6.3 AUFLS block thresholds

The Example Procurement Schedule meets the AUFLS block thresholds for most of the trading

periods in the historical dataset as follows:

2014: Block 1: 100.0%, Block 2:100.0%, Block 3: 100.0%, Block 4: 100.0%

2013: Block 1: 100.0%, Block 2:100.0%, Block 3: 100.0%, Block 4: 99.99%

2012: Block 1: 100.0%, Block 2:100.0%, Block 3: 99.98%, Block 4: 99.99%

2011: Block 1: 99.16%, Block 2:99.99%, Block 3: 99.98%, Block 4: 100.0%.

The achievement of a ‘perfect’ result depends on the availability of suitably shaped demand

units. The performance is slightly lower on the earlier 2 years that were not seen by the model

(these are the test years), partly due to a larger quantity of estimated data in those years.

Proposed method:

To meet the AUFLS block thresholds for as many trading periods as possible the selection

process uses:

a hard minimum AUFLS block threshold (absolute constraint)

a penalty cost if the maximum AUFLS block threshold is breached in any trading period

2 years of the most recent historical data to derive a sample.

Page 50: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 50 of 91

Rationale:

It is understood that from a security perspective breaches of the minimum AUFLS block

threshold are more serious than breaches of the maximum threshold. A hard minimum limit

means that the solve process must find a solution that does not violate the minimum threshold

in any trading period in the years used for the solve process.

In trials, application of a hard minimum limit was compared to application of a penalty cost for

the minimum threshold. Applying the hard limit did not noticeably increase the number of

violations to the maximum threshold.

Alternatives:

Apply penalty costs to both thresholds or hard limits to both thresholds.

When penalty costs were applied to both thresholds, even when the penalty cost applied

to the lower threshold was five times higher than the upper penalty cost, still resulted in

some breaches of the lower threshold, which is considered best avoided.

Application of a hard limit to both AUFLS block thresholds risks an infeasible result if the

solve process was not able to identify a group of demand units that can avoid any

violation of both thresholds. It also risks being less efficient as the solver may need to

select expensive demand units to meet this tight technical requirement.

Based on our trials, using two years of the dataset appears to provide a robust solution that is

not too heavily weighted in the past. The other 2 years have been used to test the solution.

Using 1 year the solution is over-fitted. It is virtually perfect for that year, but there is no

trend to identify.

Using 2 years weeds out the erratic demand units that may have been included in the

one year solution simply because they happened to be a perfectly sized puzzle piece for

one year, and not because they have a reliable profile.

Using 4 years we start noticeably restricting our available demand unit pool. Demand

units that may be a perfect fit can be excluded from the solution due to something as

simple as a single odd data issue 4 years ago. Perhaps the unit had a large load shift 4

years ago, after which it has had a reasonably consistent profile. The solver would

exclude this demand unit. It would disregard solutions with potential for very good

performance on future years because of events further in the past.

We experimented with applying weightings on years to ensure more recent periods are

weighted higher than those in the past. However this didn’t result in better solutions, instead we

just observed a similar effect as described above but less extreme.

When we look 3 and 4 years back we see much more estimated data in the trial dataset. The

more we use estimated data to select from, the less confidence we have in the results.

Page 51: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 51 of 91

6.6.4 AUFLS target levels

The mean annual target performance achieved by the Example Procurement Schedule for each

year of the four years of data are:

2014: Block 1: 10.0%, Block 2: 10.0%, Block 3: 6.13%, Block 4: 6.17%

2013: Block 1: 9.78%, Block 2: 9.75%, Block 3: 6.18%, Block 4: 6.33%

2012: Block 1: 9.78%, Block 2: 9.87%, Block 3: 5.84%, Block 4: 6.18%

2011: Block 1: 10.45%, Block 2: 9.7%, Block 3: 5.81%, Block 4: 6.02%.

To meet the AUFLS target levels the selection process applies a hard minimum constraint set at

the AUFLS block target level such that the annual net load (ANL) for the latest year of load

profile information cannot be below the target level.

The target is proposed to be applied only to the most recent year of data because this year only

is guaranteed to contain a full dataset from asset owners. A maximum target constraint is not

necessary because there is already a downwards driver on costs in the objective function.

The results are:

the ANL achieved for the reference year is at or slightly above the target

the quantity of load seen in any particular trading period may be greater or less than the

target level, and

the ANL results achieved for other historic years are seen to be within a few decimal

points both above and below the target level.

The target levels achieved are approximate rather than exactly on the 10-10-6-6. If tighter

targets are required, the impact is likely to be (a) to drive costs up, as a more exacting

requirement would result in the selection of more expensive demand units than those selected

now, (b) the selection process (running of the selection tool) will take longer, and (c) there is a

greater risk that the selection process might not be able to find a solution that meets an exact

requirement, so would be infeasible.

6.6.5 Fast response preference

‘Fast response’ is defined as the time from the relay sensing the extended contingent event to

the tripping of the demand unit circuit breaker being less than 250 milliseconds.

A small negative cost of $200 (identified as the fast response benefit rate in Schedule 3, Part 3)

is applied to each demand unit that is identified to be fast response capable.

The impact of applying the $200 benefit achieves a bias toward selection of fast response

demand units as a proportion of selected demand units, compared to the overall pool of

demand units as follows:

Total submitted fast response demand units / all demand units: 920 / 1903 = 48.3%

Page 52: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 52 of 91

Total selected fast response demand units / all demand units: 347 / 572 = 60.7%

The exact value of the benefit applied will influence the quantity selected up to a point.

Against a base case similar to the workshop 3 indicative procurement schedule, we applied net

fast response benefits of -$100 and -$1000 and achieved similar percentages of fast response

demand units in each selection: 61.9% for the -$100, and 67.4% for the -$1000. The overall

allocable costs of each selection were similar.

6.6.6 Quality level

The system operator proposes to perform a grid voltage recovery test on each proposed draft

procurement schedule. The grid voltage test models the expected voltage fluctuations occurring

at different GXPs during and after a hypothetical AUFLS event. If the voltage fluctuations at any

GXP exceed tolerances, the selection of demand units is deemed to ‘fail’ the test.

This may occur if the AUFLS supply is clumped such that too much supply is tripped in any

particular part of the network.

Given the complexity of grid voltage behaviour, it has not been possible to apply a constraint

into the model formulation to avoid this kind of problem.

Instead, the extended reserve manager proposes to respond to a poor performance report

if/when it receives one. The response will be to apply a MW constraint to the GXP or GXPs

where the selected demand unit load was too ‘clumped’ and rerun the solve process. The

constraint size applied would depend upon the circumstances of the poor performance seen.

The effect of applying the GXP constraint will be to force the model to select fewer demand

units from those GXPs and select the next most suitable demand units from other locations.

The impact was modelled on a GXP from which multiple demand units had been selected.

Applying the constraint, demand units were selected only to the constraint level (a drop of about

10 MW from this GXP). This resulted in a small increase in the overall solution cost of $270,000

and a small adjustment to the selected demand units.

No viable alternative has been identified. Initially clumping was expected to be avoided by

dividing the North Island into three regions and making three sets of selections to meet the

AUFLS block requirements for each region. With the variety of constraints in the proposed

model, the selection process runs a high risk of not achieving a feasible result in one or more

regions. The proposed methodology works well in the presence of a reasonably large dataset

but not on datasets one-third the size of the North Island. So avoiding clumping by selecting in

regional groups is not a viable option.

Q15: Do you agree that the use of averaged half-hourly historical information as a proxy to meet

the ‘at all times’ technical requirement is appropriate given currently available technology?

Page 53: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 53 of 91

6.7 Constraints applied to meet other objectives

In this section we provide a description and rationale for each of the remaining three constraints

in the selection process:

Selection cap of 60% of an asset owner’s annual average offtake.

Selection of flexible demand units.

A minimum load buffer.

6.7.1 Selection cap of 60% per asset owner

Proposal: Cap the quantity of load selected from each asset owner to 60% of their annual

average offtake (measured against the most recent submitted year of load profile information).

Rationale: While there is an efficiency/cap trade-off, providing the cap is set reasonably high

such as at 60%, the extended reserve manager considers that the benefits noted below

outweigh the costs.

We acknowledge that feedback from asset owners in workshop 3 indicated a general

preference for a lower cap (closer to 40%), though there were a range of views.13

As the efficiency benefits of the selection are best achieved with the highest possible selection

cap, we continue to propose a 60% cap.

Benefits: Capping the quantity selected per asset owner to a percentage of their annual average

offtake has these benefits:

1. Provides certainty for the asset owner, thereby promoting greater participation in the

data provision process. The minimum quantity of MW load required from each asset

owner is also set at 60% of each asset owner’s annual average offtake. The extended

reserve manager wants to encourage asset owners to submit a higher quantity into the

process where they have suitable load, as a more cost-effective result can be achieved

with a larger dataset. With a cap, asset owners can be confident that if they submit more

than the minimum requirement they will not be selected for more as a result.

2. Balances interests between extended reserve providers including limiting the difficulty

any asset owner may have in balancing their AUFLS obligation with other obligations,

such as participant rolling outage plans (PROPS). After reviewing the PROPS regime

further, we understand that the priority is AUFLS and once a selection is made, if

necessary affected asset owners would have to review their PROPS arrangements.

13 Minutes to all of the workshops are on the extended reserve manager’s project webpage at

http://www.nzxgroup.com/who-we-are/business-overview/nzx-energy/consultations-submissions/extended-reserve

Page 54: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 54 of 91

Costs: Application of a cap has a cost in the first instance on the efficiency of the selection

result, and in the second instance (if set very low) on the technical performance.

The efficiency/cap trade-off is seen in the following table and graph, which shows the allocable

costs generated from a number of scenarios based on the Example Procurement Schedule.

The proposed 60% cap is itself less efficient (i.e. $120,700 more expensive) than the no-cap

scenario.

Table 6-7: Selection cap scenarios

Cap

applied

Annual selection

cost

% selected (of

NI offtake)

Number of demand

units selected

Average $/MW

selected

No cap $5,857,093 35.8% 561 6114

60%

(proposed) $5,964,535 35.9% 572 6210

55% $6,047,717 35.8% 581 6315

50% $6,134,041 35.7% 594 6422

45% $6,227,210 35.6% 608 6543

40% $6,725,218 35.8% 650 7023

Graph 6-1: Selection caps and resulting annual selection costs

5,400,000

5,600,000

5,800,000

6,000,000

6,200,000

6,400,000

6,600,000

6,800,000

40% 45% 50% 55% 60% 65% 70% 75% 80% 85% 90% 95% 100%

Page 55: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 55 of 91

The performance of all of the scenarios against the AUFLS block thresholds was good, however

violations of block thresholds increased under an N-1 analysis14 when the selection cap was

55% or below.

The following graphs illustrate the percentage spread of loads across North Island asset owners

under different caps. The 60% cap is the Example Procurement Schedule. Three asset owners

are selected for 60% and another 2 for nearly 60% (see Graph C-1 in Appendix C for more

details and to Graph C-2 for a comparison of the MW selected from each asset owner). The no-

cap scenario resulted in selection of 100% from a direct connect and up to 81% from a

distribution company’s annual average offtake, as seen in the graph below. The 45% cap

resulted in most distribution companies being selected for 40-45% of their load.

Graph 6-2: Spread of allocation as a percentage of each asset owner’s 2014 average offtake

Q16: Do you accept the proposal to select up to 60% of the average annual offtake from any

asset owner is the most cost-effective selection?

14 N-1 analysis refers to performance of the selected AUFLS supply for an AUFLS block when the largest selected

demand unit is removed from that block.

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

60% cap No cap 45% cap

Page 56: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 56 of 91

6.7.2 Selection of flexible demand units

Proposal: The extended reserve manager proposes to select at least an additional 10% of the

target level for each AUFLS block as standby flexible reserve, equivalent to c. 1% of North

Island offtake for AUFLS blocks 1 and 2, and c. 0.6% of NI offtake for AUFLS blocks 3 and 4.

This is a total of at least an additional 3.2% NI offtake.

In addition, at least 10% of the armed demand units will be identified as flexible armed reserve.

These demand units could be disarmed to join the standby group during the operational 5 year

period if the aggregate block performance exceeds the AUFLS block thresholds.

Flexible armed and flexible standby demand units are proposed to be:

selected from demand units that have the ability to arm/disarm remotely by SCADA to

minimise extended reserve provider costs associated with arming and disarming and to

enable a reasonably prompt arming/disarming (1 business day is proposed), and

less than 2 MW in size to ensure a reasonable number of demand units are in the

flexible pool, to provide more flexibility than selection of fewer larger demand units.

If the aggregate load on any AUFLS block performs poorly against the technical requirements in

any monitoring period, one or more standby demand units will be notified to arm and/or armed

flexible demand units will be notified to disarm. In this way, procurement of flexible reserve

allows for response to change without the need to run a fresh selection process.

In response to system operator feedback, the extended reserve manager has also added a

constraint to ensure the standby group of demand units is able to cover the permanent loss of

the largest armed demand unit.

Rationale: This proposal supports the Code principle (8.54H(2)(c)) of providing for flexibility, and

supports the efficient operation of extended reserve. The benefits are considered to outweigh

the costs.

Benefits:

1. Avoided cost, delay and inconvenience to all of running a fresh selection process should

AUFLS supply levels drop below the AUFLS block thresholds or target levels.

2. Flexibility in responding promptly once fresh load profile data is available to adjust

AUFLS block levels to perform better against AUFLS requirements.

Costs:

The cost of this proposal is the procurement of an additional approximately 3.6% of North Island

offtake for standby reserve. In the Example Procurement Schedule, this represents the

procurement of 81 standby demand units for approximately $678,000.

The flexible service cost is proposed not to be applied to flexible demand units, as the provider

cost to arm or disarm a remotely-controlled demand unit is understood to be negligible.

Page 57: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 57 of 91

However if after feedback a flexible service cost is applied, then this would increase the cost of

this proposal. If applied at the rate of Beca’s proposal (which included non-remotely controlled

demand units) $220 per flexible demand unit per year would be added to the selection cost.

Description of model constraint:

To select armed flexible demand units: The model selects at least 10% of the annual

average target level for each AUFLS block using the latest year of data provided15.

Where no flexible service cost is applied as proposed, all demand units who meet the

requirements for flexible reserve are placed in the flexible armed bucket (one third of

armed demand units in the Example Procurement Schedule).

If a flexible service cost is applied following feedback, the effect will be to limit the

armed flexible bucket also to the minimum necessary to achieve the 10% requirement.

To select standby flexible demand units: The model selects at least an additional 10% of the

annual average target level for each AUFLS block using the latest year.

The constraint is approximate on purpose as an exact quantity is not required and making it

more exact could result in worse performance in other areas. There is downward pressure from

the objective function to select the lowest cost possible set of standby demand units that meet

the 10% criteria.

The decision to select 10% is not based on any firm analysis. The quantity required depends on

the extent of change that may occur over 5 operational years and this is not easy to model. As

we trial the flexible solve functionality we can see that the smaller AUFLS blocks are depleted of

demand units before the larger AUFLS blocks as time goes by. Based on consultation feedback

and on further modelling if there is evidence that a higher standby percentage may be required

over time, then we may raise the percentage, e.g. to 15%, in the final methodology.

Alternatives:

No alternatives for enabling this kind of flexibility in the selection process have been identified.

15 In addition, the 10% flexible targets (armed and standby) must be met for every trading period in the sample used

by the model to derive a least cost solution.

Page 58: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 58 of 91

6.7.3 Minimum load buffer

Constraint description: A value is determined that is equal to the largest proportion of demand

unit load to region load seen in the reference year out of all selected demand units. The hard

minimum constraint for each sample period is the sum of the region load multiplied by this value

and the AUFLS block minimum value. This builds an ‘N-1’ type buffer into the minimum

threshold. The size of the buffer is determined through the solving process and is dependent on

the demand unit dataset used.

Rationale: The load buffer aims to improve the robustness of the selection to temporary

outages, enabling temporary outages of demand units to be ignored with reasonable

confidence. It aims to reduce the risk of temporary outages causing an AUFLS block violation.

Impact: It reduces the likelihood of larger demand units being selected. They are passed over in

favour of successively smaller demand units until the lowest cost solution containing a buffer is

achieved.

Benefits: Better confidence that the total load on an AUFLS block will be within the block

thresholds even when there is a temporary outage. As a result the simplified operational design

is proposed, which does not require outage notifications.

Costs: Application of the load buffer increases the selection cost to the extent that it rejects the

largest demand units, which are the cheapest due to having lowest per-MW relay costs.

Q17: Do you support the proposal to procure an additional 10-15% of extended reserve to be

standby flexible demand units and to apply the minimum load buffer, to support flexibility in

management of extended reserve?

Page 59: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 59 of 91

7 Evaluation against the Code principles

This section identifies the key aspects of the methodology that enable it to meet the principles in

clause 8.54H of the Code.

Meeting the principle of cost-effective procurement

The key inputs into the selection process correspond to the Code’s definition of the components

of cost-effective procurement, namely:

The use of standard values which are the expected average costs of providing and

administering extended reserve, and

The adoption of customer types and standard industry interruption costs and expected

interruption hours supplied by the Authority.

The selection process itself is also relevant to achievement of a cost-effective procurement by

seeking to minimise the input costs while selecting a mix of demand units that collectively will

meet the technical requirements.

Reflecting a balance of interests between potential extended reserve providers

The proposed 60% cap on the quantity of demand unit load selected from any one potential

provider addresses this principle.

Balancing extended reserve providers and the system operator interests

The default terms and conditions that an extended reserve provider must meet when providing

extended reserve relates to this principle, as these terms will be placed by the system operator

in each extended reserve provider’s statement of extended reserve obligations. This section

is discussed in the operations section.

Seeking an appropriate balance between certainty and flexibility

The key decisions supporting this principle in the methodology are:

Selection and management of flexible reserve to maintain performance against AUFLS

requirements over multiple years.

All relays to be set and tested for all AUFLS blocks settings where possible, in the

unlikely event that a demand unit may need to be called upon to switch AUFLS blocks

during an operational period.

Applying the minimum load buffer to make the solution more robust during operations.

Q18. Do you agree that the methodology is aligned to the Code principles?

Page 60: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 60 of 91

8 Extended reserve procurement schedule

template

The proposed procurement schedule template is set out as Schedule 5 of the methodology and

the example procurement schedule uses this format.

The procurement schedule will be published as a csv file. Every submitted demand unit will be

listed in the procurement schedule and identified as selected or not selected (unless information

relating to the demand unit is identified as confidential).

The procurement schedule contains the information required by clause 8.54J(7) of the Code. If

an asset owner has one or more demand units selected, they are an extended reserve provider.

If none of their demand units are selected, they are not an extended reserve provider.

The procurement schedule will be used by:

1. Asset owners – to find out which of their demand units are selected and the amount to

be paid for each demand unit.

2. System operator – to check the draft and implementation planning.

3. Clearing manager – to calculate payments due to extended reserve providers.

8.1 Publication and commercial sensitivity

The extended reserve manager is required to publish a draft procurement schedule for

consultation, and the final procurement schedule once approved by the Authority.

Information likely unreasonably to prejudice the commercial position of the person who supplied

or who is the subject of the information is not required to be published (refer clause 8.55(6) of

the Code).

The extended reserve manager has previously asked asset owners whether any information

provided was considered to be commercially sensitive. Two responses were received:

A directly connected consumer noted their unique value of lost load was commercially

sensitive. As this asset owner is not selected in the example procurement schedule, the

results of an interruption cost calculation based on this information is not published

anyway (cost information is only provided for selected demand units).

One from a distribution company, who for the trial selection process requested that all of

their demand unit information be withheld from publication in case any commercially

sensitive situations were to arise. This party’s information is withheld from the Example

Procurement Schedule.

Page 61: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 61 of 91

Once the selection methodology is approved, the extended reserve manager proposes:

In the interests of efficient operation, which includes running a transparent consultation

process, to withhold information from publication only when positively identified

situations of commercial sensitivity are notified.

The asset owner must notify the extended reserve manager of any specific information

that is commercially sensitive by email at the time of data submission (each time).

The extended reserve manager will acknowledge the notification by email.

The information and its sensitive status may be shared with the Authority and the system

operator as part of their review of the process, and with the clearing manager for

payment purposes as required by the Code.

If a procurement schedule is not published in full this will be made clear at publication.

8.2 Consultation is technical in nature and proposed to be brief

The Code requires the extended reserve manager to undertake a selection process in

accordance with the selection methodology (8.54J(1)). Consequently, the extended reserve

manager considers the scope of the consultation on a draft procurement schedule to be limited

in nature. Manual adjustments to the procurement schedule in response to requests made

during consultation would mean the schedule would cease to be aligned to the selection

methodology.

If any changes to the draft procurement schedule are considered necessary to make as a result

of feedback, the extended reserve manager would likely rerun the selection process, which may

impact on other asset owners. If provided with a different procurement schedule to that

consulted on, the Authority would be likely to require an additional consultation round (as per

clauses 11, 12, and 13 of Part 3 of Schedule 8.5 of the Code).

When deciding which demand units to submit into the selection process each asset owner must

consider the possibility that any or all of the submitted demand units may be selected, in any

combination.

The extended reserve manager has no grounds to accept counter-proposals. Requests to

substitute a selected demand unit for other demand units for minor cost-efficiencies within a

network are highly unlikely to provide sufficient justification for rerunning the selection process.

However if a potential provider could demonstrate that the combination of selected demand

units would result in an untenable situation, or if a data input error were discovered during

consultation, then there may be sufficient justification to rerun the process.

The extended reserve manager considers therefore that the primary purpose of the

procurement schedule consultation is to enable the proposed extended reserve providers to

check for any errors in the selected demand units before the schedule is signed off.

Page 62: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 62 of 91

The extended reserve manager also proposes (in Schedule 1 of the selection methodology

under information requirements) that no submitted demand units may be withdrawn from the

selection process once the deadline for data submission is passed. If withdrawals were

allowable, then withdrawal of a demand unit after a potential provider sights the draft

procurement schedule would result in delays and decrease the efficiency of the selection

process.

Given the limited nature of the draft procurement schedule consultation, the extended reserve

manager proposes that the draft procurement schedule be released for a 2-week submission

period only. The extended reserve manager seeks asset owners’ views on whether two weeks

is sufficient for your review of the procurement schedule.

Q19. Does the procurement schedule template include the information that you require?

Q20. Do you support the extended reserve manager’s proposed process for managing

commercially sensitive information?

Q21: Do you consider a 2-week consultation period for the draft procurement schedule to be

sufficient for you to provide feedback?

Page 63: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 63 of 91

9 Operational design and obligations on

extended reserve providers

9.1 Overview of the operational design

Since the 2014 Code amendment that introduced the new extended reserve scheme, the

Authority has undertaken further work to develop operational process detail necessary to fully

specify the terms and conditions in the selection methodology. The Authority presented the

results of this work at the extended reserve workshop 3 in April 2016.

Having reviewed the aggregate profile of demand units submitted by asset owners in response

to the extended reserve manager’s initial information request, the Authority has accepted that

day-to-day operations can be simplified for extended reserve providers without materially

compromising aggregate scheme performance. This has the benefit of reducing operational

costs for extended reserve providers by avoiding the need to take action to replace temporarily

unavailable demand units.

The simplified approach has been adopted upon observing the small average MW size of

selected demand units relative to the size of each AUFLS block (in MW). Consequently a loss of

any one (small) demand unit is unlikely to result in a breach of the AUFLS block thresholds,

particularly with the application of the minimum load buffer constraint (see section 6.7.3). N-1

modelling of temporary outages shows no significant impact of temporary outages on AUFLS

block thresholds.

Under the Code, a full selection process must occur at least every five years. The operational

period is the period in between selection processes when the selected demand units are in

service.

Selected extended reserve providers will install the selected demand units with the functionality

as submitted and procured under the selection methodology and operate all selected demand

units in accordance with good electricity industry practice.

The extended reserve manager will collect load profile information each quarter-year from

extended reserve providers to undertake regular monitoring of aggregate load in each AUFLS

block. This is to provide confirmation of the quantity of extended reserve on each AUFLS block

at regular intervals to manage aggregate block load drift over time.

Regular monitoring of aggregate block load performance is a key aspect of the proposed

selection methodology. The extended reserve manager will report on the monitoring results in a

periodic performance report (PPR) for the Authority (this report will be published).

The extended reserve manager will also determine whether changing the armed/disarmed

status of any of the flexible demand units can achieve a better performance result. The model

formulation for this ‘flexible solve’ is in the methodology. The flexible solve will be performed on

the latest 2 years of load data to achieve a stable, all-seasons performance indication. Flexible

demand units will be armed or disarmed as necessary to keep aggregate demand unit demands

Page 64: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 64 of 91

within AUFLS block minimum and maximum thresholds. Thus, the quarterly review process will

serve to trim expected scheme performance at regular intervals, avoiding the need for elaborate

day-to-day operational processes.

The extended reserve manager will report on the flexible solve results in the PPR.

The system operator will adjust the extended reserve schedule to reflect any changes in

armed/disarmed status of demand units and will issue updated SEROs to affected extended

reserve providers. Extended reserve providers will be required to respond to the changed

armed/disarmed status of any demand unit within 1 business day by remotely arming or

disarming as notified.

The Authority, system operator, and the extended reserve manager will each review the PPR. If

any unforeseen circumstances or impacts are observed, the Authority may initiate changes to

the selection methodology or other parts of the scheme as necessary.

If the scheme results appear to be stable over time, the monitoring period may be adjusted to a

longer period, e.g. to 6 months. The length of the monitoring period is fixed in the selection

methodology so any changes may only be made after consulting with extended reserve

providers.

9.2 Management of change over time

Load is by nature dynamic, and changes to networks also occur over time.

As the result of a network reconfiguration, an extended reserve provider may notify withdrawal

of a demand unit if it ceases to exist. A planned or unplanned withdrawal must be notified to the

system operator as early as possible so that replacement options can be identified by the

flexible solve carried out at the next quarterly review.

The methodology provides for flexibility in the management of change first and foremost by

procuring standby flexible extended reserve – demand units that are tested and ready for

service, but not armed.

Further flexibility is proposed where flexible demand units are able to be reassigned to a

different block than originally procured for.

In the unlikely event that insufficient standby demand units are available, the Authority may

initiate a limited selection process requiring the extended reserve manager to procure additional

demand units. The model formulation for this ‘top up solve’ is included in the methodology.

Page 65: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 65 of 91

9.3 Operational obligations on extended reserve providers

The default terms and conditions proposed in Schedule 2 of the methodology become part of

each extended reserve provider’s statement of extended reserve obligations (SERO) (to be

issued by the system operator under clause 8.54P of the Code).

The default terms and conditions are set in the methodology. Consequently the time for asset

owners to provide feedback on these terms is now, as part of this consultation. The diagram

illustrates that the default terms and conditions are in each selected extended reserve provider’s

procurement notice and in the SERO.

Diagram 9: Extended Reserve High level overview

Key points are presented and explained in this section. Please refer to Schedule 2 of the

methodology for the specific wording of the clauses referred to in this section.

Obligation 1: (refer clauses 2, 3, 4, 5(b), 5(c) of Schedule 2 of the Methodology)

- When preparing demand units, comply with all technical requirements and set up

the selected demand units

The technical requirements, including testing requirements, for each demand unit are specified

in the TRS. The terms and conditions require each extended reserve provider to implement the

requirements set out in the TRS. Please comment on any technical matter in response to the

TRS consultation.

Page 66: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 66 of 91

Extended reserve procurement is set out in the methodology. The demand unit characteristics

that each asset owner submits to the extended reserve manager in the unit configuration file

during the selection process are assumed to be what will be delivered. For example, a demand

unit described by the provider as provided with flexible capability via SCADA remote control is

expected to have this capability during service. Likewise a demand unit requiring a new relay is

expected to be installed with df/dt functionality as required by the methodology.

Obligation 2: (refer clause 6(a) and 6(b) and clauses 11-14)

- Arm each demand unit by 16:00 hours on the agreed start date, and ensure each

standby (unarmed) demand unit is available for service on that date.

- Submit a change request for variations to the start date if a demand unit is larger

than 1 MW or if the date varies by more than 5 business days (refer clauses 11-14).

This set of obligations relates to the demand unit coming into service under the extended

reserve regime for the first time. The dates in the system operator’s extended reserve

schedule will match the dates in each extended reserve provider’s approved implementation

plan. You have an obligation to notify of changes to those agreed dates as set out in the change

management process.

Demand units with smaller loads may come into service up to 5 business days either side of the

start date. Armed demand units larger than 1 MW annual average load must enter service at

that date. The accuracy of the start date is relied on by three service providers:

(a) The system operator relies on it to manage system security during transition. The

system operator has proposed the 1 MW size limit.

(b) The extended reserve manager relies on it for monitoring purposes.

(c) The clearing manager relies on it for commencing payments (if any).

Obligation 3: (refer clause 6(c))

- Endeavour in accordance with good electricity industry practice to maintain the

ability of each demand unit to function at all times, restoring any temporary loss

as soon as practicable.

This is the central extended reserve provision requirement. Armed demand units are expected

to be armed and available for AUFLS at all times. Temporary unavailability can affect demand

units on a planned or unplanned basis in a variety of operational circumstances. While there is

no requirement to notify or otherwise remediate a temporary outage, the requirement is to

restore AUFLS capability as soon as practicable.

(When the extended reserve manager receives load profile information for a demand unit each

quarter, where an outage had occurred, a valid zero load entry for that time period should be

observed.)

Page 67: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 67 of 91

Obligation 4: (refer clause 6(d))

- Arm or disarm a flexible demand unit upon receipt of a fresh SERO and an

updated arming or disarming date appearing in the extended reserve schedule.

Flexible demand units must be capable of remote (SCADA) arming and disarming upon

notification of a required change, so one business day of notice is proposed. An extended

reserve provider is expected to have the communications processes in place to respond by 4pm

on the business day following receipt of an arm/disarm notification. Notification will come from

the system operator via an updated SERO, which will alert you to an updated extended reserve

schedule containing a date by which a status change to one or more demand units must be

implemented.

Arming or disarming a number of flexible demand units is anticipated following the extended

reserve manager’s quarterly monitoring process. It may be possible to provide extended reserve

providers with advance warning of a likely notification date.

However, note that notifications could occur at any time to address unforeseen operational

circumstances (e.g. a notified permanent loss of a large armed demand unit). This is expected

to be a rare event and to be unlikely to occur at short notice.

Obligation 5: (refer clause 6(e))

- Disarm a demand unit on a stop date specified in the extended reserve schedule.

A stop date represents a demand unit’s permanent retirement date. Stop dates are expected to

occur:

(a) Mostly after a fresh selection process (every 5 years or so) once a new procurement

schedule is published. An extended reserve provider would be required (when filling out

a future implementation plan) to identify a stop date for any demand unit that is not re-

selected. Once confirmed by the system operator, the agreed stop date would appear on

the extended reserve schedule.

(b) Occasionally as a result of an extended reserve provider notifying the system operator of

the withdrawal of a demand unit as AUFLS due to a network reconfiguration.

Right and obligation 6: (refer clauses 7 and 8)

The extended reserve manager may withdraw a demand unit due to its permanent

loss of capability. Withdrawals must be notified as early as possible.

Permanent loss of capability causing an extended reserve provider to retire a demand unit is only accepted where a network is reconfigured and the demand unit ceases to exist.

The intention is that:

Asset owners should not be prevented from reconfiguring networks due to their

extended reserve obligation.

Page 68: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 68 of 91

A simple process is provided to enable these occasional events to be accommodated in

the extended reserve scheme where they impact on the AUFLS supply.

To avoid confusion, the following circumstances are illustrative of the types of circumstances

that do not constitute a permanent loss of capability of a demand unit:

1. A situation that results in significant change to the half-hourly load profile for a demand

unit (whether larger, smaller, or differently shaped). For example if new customers joined

the demand unit affecting the load profile, this is not grounds for withdrawal. If the load

profile changes significantly in size or profile, the quarterly performance reviews will pick

up any consequent impact on the overall AUFLS block performance and, if appropriate,

adjust aggregate block loads by arming or disarming flexible demand units. We consider

it most efficient to maintain the selected demand units where possible during the five-

year period and to manage load drift over time at the aggregate AUFLS block level.

2. Condition-based AUFLS relay (and associated equipment) maintenance, retesting or

replacement. If a relay fails a compliance test, the extended reserve provider is required

to remedy the issue and retest the demand unit to achieve compliance, for no additional

compensation.

If you consider there should be any other valid reasons that would lead to withdrawal of a

demand unit, please provide them in your feedback.

The consequent proposed obligation on the extended reserve provider is to notify the system

operator as soon as a network reconfiguration is planned or as soon as practicable if

unplanned. The system operator will enter the stop date into the extended reserve schedule

against the demand unit and issue an updated SERO. In the event of a planned network

reconfiguration, the stop date could be several months in the future.

The extended reserve provider is responsible for ensuring any withdrawal complies with the

definition. The network configuration change should be observed in its annual asset

management plan.

The intention is that an extended reserve provider notifies the system operator as early as

possible of any upcoming withdrawals, so that actions to cover any resulting deficit in the

AUFLS supply can be planned as early as possible. Note, if you don’t immediately have an

exact stop date, notify the system operator with a best-guess date and update the stop date

later.

For monitoring purposes, you must continue to submit load profile information for any retiring

demand unit until the end of the quarter in which the stop date occurs. After the end of the

quarter in which a stop date occurs, the relevant demand unit’s AUFLS obligations cease.

Page 69: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 69 of 91

Obligation 7: (refer clauses 9 and 10)

- The extended reserve provider must provide load profile information every quarter

to the extended reserve manager. At the start, a ‘catch-up’ dataset must be

provided.

The information required for periodic performance reporting is the unit load file information for all

selected demand units (armed and standby) every three months. It is the same unit load file

provided in the same way as per the selection process, but containing the most recent quarter

of load information only.

The reporting routine is proposed to be as follows:

10 business days after end of each quarter – extended reserve providers submit data

15 business days after end of each quarter – extended reserve manager provides

revision notices

25 business days after end of each quarter – providers revise data (if needed)

30 business days after end of each quarter – extended reserve manager provides PPR

to the Authority and system operator

Future deadlines will be set and advertised in the ERM calendar by the extended reserve

manager to assist with extended reserve provider planning. These deadlines must be followed.

One of the Authority’s proposed Code amendments published for consultation in August 201616

was to require asset owners to submit accurate data, which could imply that a revision process

is not required. However, in recognition of the learning process that all parties are going through

at the start of this process, we consider the revision period is important. The extended reserve

manager proposes that the accuracy requirement would only be breached if any data issues

were not corrected by the end of the revision period.

At workshop 3, some asset owners asked whether they could submit data feeds in real time

instead of as proposed here. Authority staff have advised this option is out of scope for this

project at this time.

16 Refer to http://www.ea.govt.nz/development/work-programme/risk-management/efficient-procurement-extended-

reserve/consultations/#c16167

Page 70: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 70 of 91

9.4 Rationale for the quarterly data provision requirement

This obligation to provide regular datasets to the extended reserve manager is possibly the

largest single new ‘administrative’ requirement on extended reserve providers. However

frequent monitoring of the new scheme is the key risk management tool and as such we

consider it to be very important.

Regular reviews are intended to provide assurance to all stakeholders that the extended

reserve scheme is meeting performance expectations and that any new issues and risks are

identified and managed as soon as reasonably practicable. It enables good quality data to be

available for review at the earliest opportunity.

The extended reserve scheme is a departure from the existing AUFLS scheme and how it will

perform is as yet untested. The proposed operational approach is light-handed in that temporary

outages are ignored. The methodology relies on historic and 30-minute averages of offtake

information as a proxy for an ‘at all times’ requirement to meet the future AUFLS block

requirements. As such, regular monitoring of recent historical datasets is proposed as a crucial

part of the design to manage risk.

9.4.1 The extended reserve manager will need a contiguous dataset

For running the flexible solve process, the extended reserve manager requires a contiguous

two-year historical dataset of the load profile of each selected demand unit. Load information for

period from the last date of load profile information provided for the selection process until the

start of the reporting periods needs to be supplied.

Our intention is to ask each extended reserve provider to submit this first ‘catch-up’ dataset at

the same time that you submit your implementation plan to the system operator.

The system operator also needs this information to manage the transition process (along with

information on the retiring current AUFLS systems) so it is needed by both service providers.

We propose to start the quarterly reporting straight after the ‘catch-up’ dataset. Both service

providers will use the information during transition (implementation) of the new scheme.

Although initially none of the selected demand units will be in service on the new AUFLS blocks,

the extended reserve manager will use the information to model what AUFLS block

performance would be achieved if they were in place to check for potential issues.

The Code obligation on asset owners for supplying information during transition is the proposed

clause 8.54TD Code amendment, which is proposed to enable both service providers to collect

information required for transition and to share it with each other. The extended reserve

manager will collect it and pass the relevant information to the system operator. This is

considered to be most efficient as it saves asset owners from providing the information twice.

Page 71: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 71 of 91

Page 72: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 72 of 91

Obligation 8: (refer clauses 5(a), 6(e) of Schedule 2; and also clause 8(d)(iii) of the

methodology; and Schedule 1, Part 2, item 25; and Schedule 6, clause 3(c))

- During initial set-up and testing, configure and test each relay for as many AUFLS

block settings as the relay is capable of providing, so that:

o if necessary, you are prepared to switch a flexible demand unit to a

different AUFLS block during an operational 5-year period, and

o at the next selection process, you can inform the extended reserve

manager as to which blocks the relay is pre-set and pre-tested for.

The extended reserve manager considers the proposal may have merits but also it would create

some complications and we do not know the cost implications for you. Your feedback is sought.

If this proposal is not adopted in the final methodology, the fall-back will be the base

requirement, which is the provider must configure the relay on each demand unit to the

allocated AUFLS block setting.

Description:

This proposal is to require each extended reserve provider to set and test each AUFLS relay for

as many AUFLS block settings as it can hold and that it could supply (noting only relays with

df/dt capability would set to AUFLS block 4) as part of the initial testing process.

The cost of doing this would be covered by initial testing cost, which is applied to all AUFLS

relays (existing and new) at the start of the extended reserve scheme.

In a future selection process, if a relay is pre-tested for an AUFLS block, the reconfiguration

capital cost would be applied to a relay that is allocated to a new AUFLS block. If not pre-tested

for an AUFLS block, the initial testing cost and the reconfiguration cost would be applied to a

relay if it is allocated to a different AUFLS block.

Rationale and benefits:

(a) The extended reserve manager considers scheme would gain an added ‘back-pocket’

capability that increases its flexibility and resilience to change over time. Within a five-

year operational period if the standby pool of demand units is depleted for an AUFLS

block, flexible demand units could be shifted to the depleted AUFLS block. If they are

pre-tested and configured, then shifting to another AUFLS block could be simple, cheap,

and quick.

(b) The scheme would be more efficient and incur fewer costs in future selection processes.

During a future selection process (e.g. in five years’ time), if a demand unit is pre-set to

an identified AUFLS block setting, it could be assigned to a different AUFLS block

without incurring the initial testing cost again (to be tested for the new block setting).

Page 73: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 73 of 91

Costs:

While the proposal is simple in concept it introduces complexities in communications and

information transfer.

If the proposal to allow a demand unit to be shifted between AUFLS blocks during an

operational period were adopted:

(a) The change will be communicated via the SERO and extended reserve schedule to

providers, which must be designed into the schedule.

(b) The clearing manager’s reference for payment amounts is the procurement schedule as

advised at publication and the procurement schedule would not be altered.

Consequently the interruption cost relating to customers is not proposed to be altered if

a demand unit moves AUFLS blocks during an operational period. It is not efficient to

build this service provider system change for a process that may never be used, and for

a cost difference that, given a shorter time-frame involved (less than five years) would be

relatively small.

If the proposal to use the information in the next selection process were adopted:

(a) The unit configuration file would include fields to collect the AUFLS block numbers that

the demand unit is configured to (for use in subsequent selection processes)

(b) The selection tool would be adjusted to apply the initial testing cost slightly differently.

Unknowns – capability and cost assumptions:

As this proposal was not tested in workshops or by survey, we seek asset owners’ specific

feedback on the following assumptions that the proposal relies on about relay capability and

relay costs.

1. Most existing relays and all new relays can hold more than one and probably 4 AUFLS

block settings. (Can we assume all relays can hold 4 settings? Or all new relays can

hold 4 settings?)

2. There is very little difference in cost between configuring and testing for one block

setting or more settings and the cost can be covered by the proposed initial testing cost.

3. A flexible demand unit (that can remotely arm and disarm via SCADA) and that is

already configured and tested for a second AUFLS block, can switch to that AUFLS

block via remote control, i.e. switching blocks costs no more than arming or disarming.

4. If in a new selection process a demand unit is allocated to a different AUFLS block for

which it was already configured and tested, no more testing would be required. However

some demand units would require a site visit to make the switch. (The configuration cost

would be paid upon allocation to a new AUFLS block.)

Page 74: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 74 of 91

Alternative:

The alternative is to not attempt to add this capability to the scheme. The basic requirement is

that each relay must be set to the allocated AUFLS block setting. This alternative is simpler to

implement and is the minimum necessary for the scheme’s operation.

However, the possible efficiency and flexibility benefits are not achieved. The next available

management option if standby demand units are depleted is to initiate a ‘limited selection solve’,

which is a more costly option.

We seek your specific feedback on this proposal in order to decide whether and how to include

it in the final methodology.

Q22. Do you support the proposed operational design of the extended reserve scheme?

Q23: Do you have any comments on the flexible solve and limited selection processes

described in sections 5.4 and 5.5?

Q24: Regarding Obligations 1 to 7: Do you have any comments or feedback, for example on the

information requirements and proposed timeframes?

Q25: Regarding Obligation 8 (the proposal to require all relays to be set to and tested for as

many AUFLS block settings as the relay can supply):

a. Are the capability and cost assumptions correct? Please comment on any that are not.

b. Do you support the proposal to require all relays to be set to as many AUFLS block settings

as they can hold and supply?

c. Do you support the proposal to require flexible demand units to switch between AUFLS

blocks during an operational period, if necessary to improve flexibility?

Page 75: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 75 of 91

10 Compensation payments for extended reserve

The extended reserve manager proposes in the methodology to introduce a payment

mechanism for extended reserve (clause 14). Schedule 6 of the methodology sets out how the

payments would be calculated. In this section we explain the rationale for making this proposal.

10.1 The Code and asset owner feedback to date

Feedback from asset owners from industry workshop 2 on a payment mechanism was:

Views for Views against

There needs to be a

payment mechanism

to get rid of free

riding.

If there is no

payment mechanism

the asset owners

who do not provide

AUFLS won’t face

any costs, but those

who do provide

AUFLS will face the

cost of the

inconvenience of

AUFLS interruptions.

Without a price that

is paid you cannot do

equivalence with

other options.

Payment should be

based on the

marginal cost of Voll.

It is not practical to transparently pass costs and payments through

to those who benefit and those who suffer costs by being shed.

By accepting payment we incur risk of liability and compensation

risk.

Payments add a whole level of complexity and costs which are not,

in our opinion, justified.

Overhead costs of a payment mechanism are too high.

We are a small company and don’t want the administrative hassle.

The reason for having extended reserve is to avoid a complete

collapse of the NI grid. This should be done on the most efficient

basis, not on the basis of a ‘market’.

Requirements should be mandated on both distribution companies

and direct connects.

Trust owned network companies have consultation obligations to

their beneficiaries (customers), this hasn’t been allowed for.

Needs to be a cost recovery mechanism for real costs, but the

proposed construct is too generic and doesn’t necessarily reflect

individual costs. Where it creates a new cost to consumers it is less

desirable as increases cost to serve.

Consideration that trust-owned EDBs have different obligations and

are not always seeking to pass through costs so the assumption in

the model may not hold for all.

The Authority has previously consulted on and codified the requirements for the extended

reserve scheme and decided that if there is to be a compensation payment as part of the

extended reserve scheme then the beneficiaries of the scheme will pay17. The beneficiaries of

the scheme have been defined in the Code (8.67A) as connected asset owners who are further

defined to mean a direct consumer or a distributor in its capacity as the owner or operator of a

local network.

17 The extended reserve manager understands that the Authority and the Commerce Commission have come to an

understanding on the pass through of extended reserve costs to consumers. For specific advice on the issue please

contact the Authority or the Commerce Commission.

Page 76: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 76 of 91

Beneficiaries would be charged ‘allocable costs’ only, which are the costs incurred by extended

reserve providers. The extended reserve manager and system operator costs for developing

and administering this scheme are not ‘allocable costs’ to be paid by beneficiaries but are

included in the Electricity Authority’s budget allocation for service providers (levy funded).

The Code 8.54G requires the extended reserve manager to prepare a methodology that must

specify how payments to asset owners for providing extended reserve are set. In addition the

Code (8.54H) requires the extended reserve manager to set out how it will enable extended

reserve to be procured cost effectively. Schedule 6 of the methodology sets how the extended

reserve manager proposes to determine how the payments to asset owners are set.

The clearing manager is required by 8.67A to allocate the net cost/payment (if any) to/from the

connected asset owners.

10.2 Options for a compensation payment mechanism

The extended reserve manager considers that there are three high level options for a

compensation payment regime that will meet the principles set out in clause 8.54H of the Code

which are:

a. A compensation payment mechanism as per the draft selection methodology (based on

generic relay costs and potential interruption costs).

b. A compensation payment mechanism as per the selection methodology but with

compensation rates based on actual asset owner installation and operational costs.

c. No compensation payment mechanism.

The make-up of sub-options for (a) above can have many alternative categories and rates. To

meet the principles of 8.54H(b) if there is to be a payment compensation mechanism it must as

a minimum include the expected cost of providing the extended reserve, the expected costs of

an interruption, and the likely transaction costs associated with administering and providing

extended reserve.

Option (b) has the benefit of discovering the actual installation and operational costs of the

scheme but would be administratively unmanageable as it would require audit of the proposed

costs of each demand unit offered. This option is not discussed further.

The Option (c), of not having a compensation regime, still meets the principles providing the

selection process still selects demand units utilising the generic cost categories and rates set

out in the selection methodology.

Page 77: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 77 of 91

10.3 What is the policy intent of having a payment mechanism?

The policy intent of including a compensation payment mechanism is set out in a paper18

produced by the Authority. The policy intent is:

1. to incentivise the provision of enhanced AUFLS services

2. to incentivise direct connect consumers to submit realistic Volls for the selection

process, and

3. to equitably spread the burden of AUFLS provision across all consumers.

10.3.1 Can introducing a payment mechanism achieve this policy intent?

We review each policy intent in light of the experiences we have had through the trial data

collection, feedback from with asset owners to date, and the proposed methodology.

1. The payment mechanism offers limited if any incentives for enhancing AUFLS

services.

Three possible enhancements have been considered: faster demand unit response times, df/dt

functionality, and remotely-controllable arming/disarming capability. Based on the trial data

received, asset owner feedback and the Beca report, the extended reserve manager proposes:

that a fast response incentive payment is removed. In many cases, a more modern

circuit breaker must be installed to achieve a faster response time19. A significant

payment would be required to incentivise faster response times with consequential

impact on the AUFLS system costs

to mandate that all new relays are provided with df/dt capability. Most modern digital

relays that would be used for extended reserve already include this capability and it is

more efficient to enable this capability with installation of a new relay and regard all new

relays as having this capability in the selection process. The cost of the df/dt capability is

included in the base relay installation cost; and

that all flexible demand units are only selected from demand units that are already

offered with remote arming/disarming capabilities. A flexible service payment is

proposed not to be included on the basis that the actual cost of remotely arming is

negligible. However we request feedback on whether including a modest payment would

incentivise others to introduce remote arming capability on more demand units.

18 Discussion on Extended Reserve Payments Mechanism 3 November 2015, available at:

http://www.ea.govt.nz/development/work-programme/risk-management/efficient-procurement-extended-

reserve/events/papers-from-extended-reserve-workshop-2/

19 A fast response conversion cost was calculated to be $27,000 in the report prepared by Beca for the extended

reserve manager, 7 April 2016. http://www.nzxgroup.com/who-we-are/business-overview/nzx-energy/consultations-

submissions/extended-reserve

Page 78: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 78 of 91

Thus the ability for the payment mechanism to incentivise enhanced services is at best limited,

and not sufficient on its own to justify introduction of a payment mechanism.

2. Incentivising direct connects to submit realistic Volls is not required.

Some parties were concerned that directly connected consumers would attempt to ‘exempt’

themselves from selection by submitting unrealistically high Volls for their load. In the trial data

process we found the directly connected customer Volls submitted were realistic, and on

average (excluding one) were slightly less than $11,500/MWh20. If substantially different Volls

are submitted from these parties in the ‘real’ data process, they will be challenged. Therefore

this reason on its own is not sufficient to introduce a payment mechanism.

3. The payment mechanism is required to equitably spread the burden of AUFLS

provision across all consumers.

In the extended reserve manager’s view this policy intent is only achieved with a payment

mechanism.

From a provider perspective, the payment mechanism does introduce additional complexity and

cost, of an average of $4,700 each provider/beneficiary, or across parties, up to $110,000 per

year21. From a provider perspective also the lower the selection cap on the percentage of

demand units selected per provider the more equitable the spread as each provider is having to

provide a similar portion.

From a consumer22 perspective, without a payment mechanism the burden of costs from the

actual provision of extended reserve falls unevenly. Consumers serviced by distribution

companies who have a proportionately high allocation will pay more than consumers serviced

by distribution companies or directly connected consumers with a low allocation. In the Example

Procurement Schedule 5 of the 7 direct connects will effectively be ‘exempt’ from the extended

reserve scheme in the absence of a payment mechanism, and all benefit disproportionately.

This is one of the issues that the extended reserve manager understands the Authority wanted

to address.

The extended reserve manager could allocate more evenly across providers by placing the

lowest possible selection cap (say, 40%) on the process. However this increases the cost of the

scheme considerably as a less efficient selection is made. In the Example Procurement

Schedule the selection cost is $5.965 million per annum, with a 40% cap it is $6.725 million or

20 This figure excludes one high value submitted for one direct connect (which is also withheld due to commercial

sensitivity).

21 Beca’s report gathered the internal provider costs of dealing with the clearing manager’s invoices. The average

cost surveyed was $4,700 per provider/beneficiary.

22 Distribution companies can pass their extended reserve costs to consumers and direct connects are also

consumers.

Page 79: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 79 of 91

$760,000 more. Even when the cost of dealing with the clearing manager ($110,000 per

annum) is added to the Example Procurement Schedule cost, the Example Procurement

Schedule selection outcome is more efficient by $650,000 per annum.

Consequently the extended reserve manager considers it is in the long term benefit of

consumers to spread the cost of the scheme across all North Island consumers, despite the

additional cost of managing payments. From a national perspective a more efficient selection of

demand units is possible while providing for equity through payments so that the extended

reserve service is provided for a lower net cost overall.

10.3.2 Incentives and payments

Prudent asset owners will act to maintain and improve their asset base for least cost. The

presence or absence of a payment mechanism may alter asset owner incentives and decisions.

The net allocation cost position of each asset owner relates only partly to the quantity of

extended reserve they provide. It also relates to the relative average size of their demand units

and the specific costs of their demand units. A provider can be selected for 60% of their offtake

and still be a net payer. Such asset owners may be incentivised to alter their data submission

where possible. The resulting potential impact on all parties would be to potentially increase

cost. Refer to section 6.4.1 for more detail.

With a payment mechanism the extended reserve manager considers that in general asset

owners may be:

more incentivised to submit demand units that are likely to be selected, to offset their

costs with a payment

more likely to submit demand units with new relays, as the cost is spread, and

less likely to submit a lot of very large demand units, as the payment received is

proportionally less than for several small demand units.

Without a payment mechanism, the extended reserve manager considers that assets owners

may be:

more likely to attempt to avoid selection by submitting less attractive demand units, as

they cannot offset their costs of provision if selected, and

more likely to submit existing relays.

Either way, the extended reserve manager considers that asset owners are most likely to

submit a set of demand units that reflects their current AUFLS provision and an additional set to

meet the required quantity of data, that least impacts on their overall business cost and on their

customers.

Q26. Do you support the proposal to introduce a payment mechanism and why?

Q27: Can you identify incentives including perverse incentives or ‘gaming’ opportunities in the

extended reserve selection process, whether there is a payment mechanism or not?

Page 80: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 80 of 91

10.4 Methodology for calculating the payment

To satisfy the principles set out in the clause 8.54H of the Code the methodology sets out the

values on which a cost-effective selection is made, which is also the basis on which payments

are proposed to be calculated (in Schedule 6). It is intended to be as simple as possible.

There are three categories of proposed compensation payments:

Payment to compensate customers for having their load interrupted:

o Demand unit interruption payment (DUIP)

Payments for the provision of AUFLS assets:

o AUFLS system administration payment (ASAP)

o Relay operating payment (ROP)

o Relay capital payment (RCP)

Payments for the provision of enhanced services:

o None are proposed but feedback is sought on whether a flexible service cost

could be an incentive payment.

The compensation payment categories, calculation mechanism and rates are discussed in the

AUFLS provision costs section 6.1 of this paper.

The arithmetic sum of these payment categories for the selected demand units per provider

forms the standard annual amount payable to each extended reserve provider. The clearing

manager will divide this annual payment by 12 to allocate it to the provider for each billing

period.

The methodology proposes that the clearing manager pay the interruption payment (DUIP) only

for those demand units that are armed in each billing period (and not pay standby demand

units). This would save some $269,000 from the Example Procurement Schedule annual

selection cost. The rationale is customers are only at risk of being interrupted when the relay is

armed23.

Q28. On the assumption that there is a compensation payment regime do you agree with the proposed compensation details?

Q29: Do you have any other comments to make on the proposed methodology?

23 The clearing manager will assume the DUIP is paid by the number of days in the month that it is armed. The

clearing manager will assume the other three charges (ASAP, ROP, and RCP) are paid for the full month from the

month of the start date.

Page 81: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 81 of 91

Appendix A: Draft Selection Methodology

Please refer to the pdf provided in this consultation pack.

Page 82: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 82 of 91

Appendix B: Example Procurement Schedule

Please refer to the Excel file, ‘Example Procurement Schedule – 1 Sep 2016’ provided in this consultation pack.

Note one party’s information is withheld due to notification of potential commercial sensitivity.

Page 83: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 83 of 91

Appendix C: Analysis of Example Procurement

Schedule

Table C-1: Quantity of selected demand units

Block

# demand units

(armed

inflexible)

# demand units

(armed, flexible)

Total # armed

demand units

# demand units

on standby

Total #

selected

demand units

1 127 61 188 26 214

2 77 61 138 27 165

3 59 25 84 13 97

4 58 23 81 15 96

Totals 321 170 491 81 572

A third of the armed demand units are identified as armed flexible, in total representing 7.6% of

the total North Island annual offtake.

Table C-2: Quantity of selected North Island offtake (MW)

Block MW armed % NI offtake

armed MW standby

% NI offtake

on standby

Total % NI

offtake selected

1 268 10.0% 28 1.1% 11.1%

2 268 10.0% 28 1.1% 11.1%

3 162 6.07% 19 0.7% 6.77%

4 167 6.25% 20 0.7% 6.95%

Totals 865 32.32% 95 3.6% 35.92%

Table C-3: Allocable cost breakdown

Example procurement schedule cost breakdown Totals %

Total interruption cost $2,551,510 42.8%

AUFLS provision cost $3,413,025 57.2%

Total allocable cost (selection cost) $5,964,535 100%

Page 84: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 84 of 91

Table C-4: Net indicative cost allocation per asset owner

Example procurement schedule results under the draft selection methodology

Indicative annual allocation of costs to beneficiaries24

Asset owner

Number of demand units selected

MW selected

% selected of own annual offtake

Receives as ER provider (annual $)

Allocation proportion (% of 2014/15 NI offtake)

Pays as beneficiary (annual $)

Net $ to (receive) or pay

Unison 68 80.48 51% 743,199 5.87% 350,206 -392,994

Electra 18 22.42 60% 201,331 1.41% 83,869 -117,462

Waipa Networks

15 25.30 59% 196,669 1.61% 95,902 -100,767

Horizon 14 31.24 60% 186,835 1.95% 116,490 -70,345

WEL Networks

38 50.51 47% 306,882 4.03% 240,645 -66,236

The Lines Co 8 13.30 50% 114,030 1.00% 59,627 -54,404

Top Energy 11 10.27 55% 88,147 0.70% 41,511 -46,636

Counties Power

14 20.88 35% 180,558 2.25% 134,026 -46,562

Eastland Energy

3 19.23 60% 112,165 1.21% 71,904 -40,261

Scanpower 3 4.19 44% 36,867 0.35% 21,056 -15,810

Northpower 33 42.59 37% 265,724 4.32% 257,624 -8,099

Other 0 0 0% 0 0.08% 4,641 4,641

Kiwirail Holdings

0 0 0% 0 0.18% 10,874 10,874

Methanex 0 0 0% 0 0.23% 13,503 13,503

Kupe Gas Project

0 0 0% 0 0.27% 15,909 15,909

Centralines 0 0 0% 0 0.49% 29,234 29,234

PowerCo 123 156.12 30% 1,093,868 19.14% 1,141,350 47,482

Winstone Pulp

0 0 0% 0 1.02% 60,718 60,718

Norske Skog 2 6 11% 45,624 2.07% 123,219 77,595

PanPac 1 2.16 4% 21,575 2.23% 132,867 111,292

Wellington Electricity

21 158.97 60% 444,865 9.97% 594,639 149,774

NZ Steel 0 0 0% 0 3.03% 180,551 180,551

Vector 200 316.89 32% 1,926,167 36.62% 2,184,169 258,002

Totals 572 961 36% 5,964,53525 100% 5,964,535 0

24 The allocation of costs to beneficiaries is determined by clause 8.67 of the Code. The cost allocation is not

determined by the methodology so is out of scope of this consultation process.

25 This is the maximum possible payment assuming no demand units are on standby. The annual payment would be

approximately $269,000 lower as it is proposed to not pay standby demand units the interruption payment (or DUIP).

Page 85: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 85 of 91

Graph C-1: Percentage selected of asset owner’s 2014 average offtake

Graph C-2: Selected quantity (MW) per asset owner

0%

10%

20%

30%

40%

50%

60%

70%

0

50

100

150

200

250

300

350

Page 86: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 86 of 91

Table C-5: MW selected from distribution companies & direct connects

2014/15

offtake (MW)

% of total

NI offtake

MW

submitted

% of total MW

submitted

MW

selected

% of total MW

selected

Distribution

companies 2434 91% 1976 98% 953 99%

Directly

connected

consumers

243 9% 38 2% 8 1%

Totals 2677 2014 961

Table C-6: Demand units selected from distribution companies & direct connects

# DUs submitted % of total

DUs offered # DUs selected

% of total DUs

selected

Distribution

companies 1878 99% 569 99%

Directly connected

consumers 15 1% 3 1%

Totals 1893 572

Table C-7: Load-weighted proportions of submitted and selected customer types

Customer type Load-weighted %

submitted

Load-weighted %

selected

Total MW

selected

Light industrial and primary industry

($8,200 interruption cost (IC)) 10.4% 14.6% 141

Residential ($15,900 IC) 45.9% 46.2% 444

Heavy industrial ($19,500 IC) 6.5% 7.9% 76

Commercial ($37,200 IC) 29.1% 25.5% 245

Public health & safety ($100,000 IC) 3.0% 2.3% 22

Large User (variable IC) 5.1% 3.4% 33

Page 87: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 87 of 91

C.1 Selection of public health and safety customers

11 submitted demand units (27.7 MW) with 100% of the public health and safety

customer type were excluded from the selection process as proposed in the

methodology.

The example procurement schedule contains 273 demand units with some quantity of

public health and safety in them, and of those, 194 demand units are on the current

AUFLS scheme.

C.2 Other statistics

Number of existing relays selected: 394. Number of new relays selected: 178.

Number of selected demand units currently on AUFLS compared to the total number

selected: 394 / 572 (69%)

Number of demand units selected that are fast response demand units: 347 / 572 (60%

of total selected) compared to the submitted quantity of 917 demand units out of 1893

(48%).

Number of Transpower-owned relays selected: 32.

C.3 Selected demand unit size

Average size of selected demand units: 1.68 MW (average submitted size: 1.10 MW)

Largest selected demand unit: 13.2 MW

Smallest selected demand unit: 0.011 MW.

Table C-8: Percentage selected compared to size

Demand unit size Quantity submitted Quantity selected Percentage

selected

> 5 MW 31 25 81%

1 MW – 4.99 MW 761 400 53%

0.5 MW – 0.99 MW 616 140 23%

< 0.5 MW 485 7 1.4%

The selection results show that, all things being equal, larger demand units are favoured over

smaller demand units. This is because they are cheaper due to the per-relay AUFLS provision

cost component, which is a fixed cost within the demand unit.

Page 88: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 88 of 91

Appendix D: Data specification

Please refer to the separate document provided in this consultation pack.

Page 89: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 89 of 91

Appendix E: Draft of Part 6 of Schedule 4

Prepared in response to the system operator’s comments on the draft methodology.

Page 90: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 90 of 91

Page 91: Consultation Paper...Electricity Industry Participation Code 2010 (Code), which with the operational design approach developed by the Electricity Authority (Authority), constitute

CONSULTATION PAPER – DRAFT EXTENDED RESERVE SELECTION METHODOLOGY – 11 Oct 2016 91 of 91

Appendix F: Questions

Please refer to the Word document for a table of consultation questions provided for submitters.