Document Actions
October 2007 revision to draft grid planning assumptions (GPAs) –generation scenarios
Index
Background
In September 2006, the Commission developed draft grid planning assumptions (GPAs) as an initial step in preparing the 2007 Statement of Opportunities (SOO). This included the preparation and publication of demand forecasts and four generation scenarios for consultation with interested parties.
In May 2007, the Commission decided to delay the publication of the next SOO until the NZ Energy Strategy was finalised.
Now that the Government’s climate change policies have been developed further and the New Zealand Energy Strategy has been finalised, the Commission is recommencing work on preparing the next SOO, which is likely to be published in 2008. As an initial step, and following the receipt of additional information from Transpower, the Commission has revised the generation scenarios and other key inputs to the draft GPAs that were released in May 2007. This page contains those revisions.
The Commission and Transpower have also been discussing the draft GPAs published in May 2007 in the context of identifying scenarios to be used in applying the grid investment test (GIT) to a number of investment proposals, including an anticipated HVDC investment proposal.
In particular, the Commission is considering a "90% renewables by 2025" scenario which models the NZ Energy Strategy target of 90% of electrical energy produced from renewable resources by 2025. That scenario would be additional to the four scenarios anticipated by the draft GPAs. The weightings of the four draft GPAs released in May 2007 would therefore be affected.
The Commission is also considering revised assumptions regarding peak capacity requirements when modelling possible future generation developments in the scenarios anticipated by the draft GPAs released in May 2007.
More detailed work will be carried out over the next few months working towards a publication of the SOO in 2008, beginning with the release of an updated demand forecast.
Contact Brian Bull for any further information on the Commission's work programme on generation scenario development.
Generation scenarios - Outline
This section provides a high-level description of the five scenarios currently under development. More detailed descriptions of the scenarios are given at the bottom of the page.
Generation expansion methodology
Modelling approach
In order to develop each scenario with a reasonable degree of internal consistency, the Commission has developed a Generation Expansion Model (GEM). GEM is a mixed-integer linear programming model and is used to find a least-cost schedule of the timing and location of new generation capacity, subject to energy and capacity constraints. GEM's data inputs include:
The generation scenarios are cross-checked using SDDP (a commercial medium-term hydrothermal dispatch model; working SDDP input files), with a particular focus on unserved energy, the use of hydro storage and HVDC flows. They will also be checked using the Commission's in-development capacity model.
A notable feature of all the scenarios is the extent to which capacity is added to meet peak demand requirements rather than to meet overall energy requirements (resulting in, for example, the introduction of increased demand-side participation, thermal peaking generation and eventually pumped hydro). These outcomes reflect a potential transition of the New Zealand power system from being "energy constrained" to facing both energy and capacity constraints. Contributors to this change include the emphasis on base-load plant over the last 15 years and the amounts of new baseload and intermittent plant being added in the scenarios. The GEM model includes explicit capacity constraints, which have a major influence on the development of the scenarios. Previously these capacity constraints were designed to require that N-2 security be achievable at cold-year winter peak demand. However, the Modelling Group has reconfigured the GEM model to use a N-1 standard, which is considered to be a more reasonable assumption.
The Commission is aware of some comments to the effect that it is not clear that the current market design would lead to a level of generation investment that met the modelled capacity constraints. Other Commission work programmes are currently assessing this issue.
While the particular assumptions adopted in relation to a specific investment proposal will be considered by the Commission in the course of carrying out its decision-making responsibilities under part F, for the purposes of continuing the Modelling Group’s work the Commission will continue to assume that security of supply standards will be maintained by some means. Similarly, all scenarios will include enough generation to supply a 1-in-60 dry year (given a reasonable level of demand response).
Generating technologies
GEM models a wide range of large-scale electricity generation technologies. The main technologies included in the scenarios are:
Demand-side measures are modelled as another 'generation technology'. Three main types of demand response are included in the scenarios:
The scenarios assume fuel prices of $4/GJ for black coal, $1.80/GJ for lignite, and $25/GJ for diesel. Gas prices are assumed to increase over time, in response to diminishing supply and (in some cases) exposure to international markets. The gas price rises from $5.5/GJ initially to $7/GJ in 2012 in all scenarios. It continues to rise to $10/GJ by 2020 with a limit of 75 PJ/year, except in the 'High Gas Discovery' scenario, where the price remains at $7/GJ with a limit of 120 PJ/year, and the 'Mixed Technologies' scenario, where the construction of an LNG terminal in 2020 allows 'unlimited' LNG imports at a price of $10.50/GJ.
For some plant (notably gas generators in or north of Auckland) fuel transport costs are included.
Carbon costs are additional to the fuel prices above. GEM models the cost of carbon relatively simplistically, as a fixed cost per tonne CO2-equivalent. This cost is introduced at a low level in 2010, rises to a maximum level in 2018 and remains constant thereafter. The maximum level is $15/t CO2-equiv for the Mixed Technologies and High Gas Discovery scenarios and $40/t CO2-equiv for all the Renewable scenarios.
The carbon costs assumed in GEM predate the Government's announcement of the Emissions Trading Scheme and may need review. However, since the actual cost of carbon under an Emissions Trading Scheme (ETS) would be determined by the market and could vary from year to year, it may not be possible to produce more accurate estimates at this stage.
These fuel and emission costs are also given in the GEM data spreadsheet, which is distributed with the GEM model.
Modelled generation projects
The scenarios are built based on lists of generating plants. These fall into into three categories:
A key input to this list of projects was the following report by PBA, providing information on existing and potential thermal, geothermal, wind, and hydro schemes.
Readers should note that the inclusion of a project on the list does not necessarily mean that any developer is considering that project. In fact, some projects on the list have been specifically disavowed by generators. Nonetheless, they are included for completeness and in case they are taken up by developers in the future.
The complete list of possible future plants in the model is included in the GEM data spreadsheet, which is distributed with the GEM model. The spreadsheet includes assumptions on project capacity, cost structure, resource quality, and earliest possible commissioning date.
Many of these generation assumptions have changed since the publication of draft GPAs in May 2007. They are potentially contentious, and the Commission welcomes feedback or further information.
At time of writing, the list of projects includes:
Note that GEM does not model relatively small embedded plant (such as backup generators, very small hydro schemes, and distributed wind generation). The electricity demand forecasts implicitly assume that the contribution from embedded plant is maintained at current proportions, which would require the gradual addition of distributed generation.
Retirements
GEM does not model plant retirements endogenously; these must be provided to the model as input assumptions. Currently, all new projects and most existing stations are assumed to remain in operation until 2042 at least. The exceptions are;
We acknowledge that all these retirement assumptions are highly speculative. In fact, the owners of these plants have not stated when they plan to retire them.
Energy constraints
GEM requires that scenarios include enough generation to supply demand in a dry year, given modelled levels of demand-side response. Further generation can be added on an economic basis and/or to satisfy the capacity constraints.
The build schedules in the scenarios are optimised using a two-step process. Firstly, project build times are selected on the basis of optimality (minimum discounted post-tax cost) in a model run where hydro inflows are assumed to be 97% of the historical average. Secondly, thermal peakers can be introduced or brought forward on the basis of optimality in a model run with dry-year (1932) hydroelectric production. The intention of this process is to model a future in which baseload is constructed to meet normal levels of need, but additional thermal peakers can be constructed for dry-year reserve (or to meet unexpectedly high levels of demand growth).
(Dispatch simulations are then carried out using a range of hydro inflow scenarios, as discussed in the GEM documentation.)
GEM models electricity demand via load duration curves. Currently these load duration curves are quarterly, with four load blocks per quarter (though GEM will soon allow more flexibility here). The total annual demand in each island is based on the draft GPA demand forecast, with allowances for AC losses and to avoid double-counting of modelled embedded generation. The shape of the load duration curve is based on a historical reference curve.
The demand forecast is included in the GEM data spreadsheet, which is distributed with the GEM model.
Electricity production from existing hydro plant is estimated outside the GEM model. The SDDP model was used to estimate monthly electricity output from the existing hydro system, under a range of historical inflow sequences from 1932 to 2004. These production figures have been pooled for each island and entered into GEM as 'scheduled hydro'. The assumption is that existing hydro plant will continue to operate much as it has in the past. Hydro storage from one quarter to another is not explicitly modelled in GEM (production cannot be carried over to a subsequent period) – however, the interperiod storage dynamics were modelled in SDDP and are incorporated in the production figures.
Capacity constraints
GEM includes three capacity constraints; these are 'hard' constraints which must be satisfied by all build schedules. They require a 'N-1 at cold-year winter peak' standard – in other words, generation must have the technical capability to serve the maximum likely winter peak demand in a cold year, with enough margin to cover a single contingency. The three constraints each model a different contingency:
These constraints affect the build schedule in several ways. They bring in demand-side options; later in the scenarios they lead to the construction of thermal peakers and/or pumped and peaking hydro; and they add an incentive to favor mid-order plant over baseload. All these effects apply more strongly in scenarios with significant amounts of intermittent generation (wind, wave and/or run-of-river hydro), and they would also apply more strongly in scenarios with less HVDC capacity.
The peak demand figures used in the forecasts are derived from the annual half-hourly peak demand forecasts published as part of the draft GPAs from May 2007. The GEM peak forecast was produced by starting from the Commission's prudent forecast from 2007 and growing it forward at the growth rate of the expected forecast. This was intended to predict cold-year demands under a scenario of average underlying demand growth (as opposed to using the prudent forecast throughout, which would model a scenario of high underlying demand growth). The half-hourly forecast was then increased by 90 MW (national) or 60 MW (North Island) to allow for within-half-hour variation. An allowance for AC losses was added on, assuming AC transmission losses at peak of 4.0% in 2007, trending upwards to 5.5% in 2030 due to increasing system loads.
A peak contribution factor has been specified for each technology, for use in these constraints.
In all scenarios, carbon charges have the effect of encouraging the production of electricity from renewable sources.
The '90% Renewables by 2025' scenario also models an explicit requirement for 90% renewable electricity by 2025. The required renewable percentage increases linearly from 70% in 2010 to 90% in 2025 and remains at 90% thereafter. This is modelled as an average-year requirement, not a dry-year requirement – the renewable percentage is allowed to dip below 90% in dry years, but is over 90% in wet years and averages out to approximately 90% in the long run.
For this purpose, 'renewable' technologies are deemed to include geothermal, wind, hydro, marine, and coal with carbon storage and sequestration. Other forms of thermal generation are still used for peaking and balancing, but do not count as 'renewable'.
WACC and discount rate
The generator WACC assumed in developing the generation scenarios is 8% real post-tax. This is slightly higher than the earlier assumption of 7.7%, to notionally account for hurdle rate issues as well as to make it distinctly different from the 7% assumed for transmission investments.
Concept Consulting has prepared advice for the Commission on this issue.
Note that GEM can be instructed to process results using different discount rates such as 4%, 7% or 10%, but these should not be confused with the 8% generator WACC used in the objective function.
Locational effects
Location factors have been used to reflect the effects of locational price on project economics. Each modelled project is mapped to a transmission region, and each region has a location factor associated with it. For example, an identical project would require a lower average price to be constructed were it located at Otahuhu than it would were it located at Tiwai. The location factors used are constant and do not change in response to generation and demand changes.
Note that cost outputs can be produced exclusive of these location factors in the current version of GEM (unlike earlier versions).
AC transmission and distribution losses are built into the demand forecasts used in GEM, using island-level loss figures (rather than being based on the location factors above). Losses on the HVDC link are modelled explicitly using a piecewise linear approximation to the loss function.
Capital cost uncertainties
There is considerable uncertainty around the capital costs and potential benefits of the modelled wind and hydro generation projects, which is not reflected in the baseline project cost/benefit estimates in the list of modelled projects. As a consequence, the unmodified cost curves for wind (and hydro, to a lesser extent) are very flat, with relatively little variation in LRMC between projects. In reality these curves should be more sloping, with some potential projects delivering more or less benefits than shown, or able to be constructed at a lower or higher capital cost.
In the scenario development, this is implemented as follows. Modelled wind and hydro projects (excluding committed and 'highly likely') are divided into mutually exclusive subsets of roughly equal size, on an arbitrary basis. In each scenario, one of these subsets is selected, and it is assumed that projects in this subset possess unexpected advantages (e.g. can be constructed more cheaply, or can produce more energy, than the baseline predictions indicate). These advantages are modelled by reducing the capital costs of the selected projects by 10% for hydro, or 20% for wind (where the uncertainties are considered to be greater). The effect is to shift the selected projects higher up the merit order. This has the effect of increasing the slope of the cost curves for wind and hydro, with the affected projects becoming more competitive with other forms of generation (i.e. thermals). A useful secondary effect is to increase the diversity between scenarios - each scenario has a different set of projects with reduced costs, and hence a different build order.
Transmission charges
South Island generation projects are assumed to pay a charge of up to $40/kW p.a. for the HVDC. The figure of $40/kW was derived by annualising estimates of Transpower's future HVDC charges and apportioning them to an assumed South Island installed capacity. (This figure will be reviewed soon in light of indicative cost recovery figures provided by Transpower.)
For generators who already have significant installed capacity in the South Island, the marginal increase in HVDC cost allocation per MW of new generation is less than $40/kW. (Consider a new plant with capacity equal to 1% of installed South Island capacity. If this plant belongs to a new entrant generator, then their share of total capacity rises from 0/100 to 1/101, i.e. to almost 1%. If, on the other hand, it belongs to a generator that already has 70% of island capacity, then their share of total capacity rises from 70/100 to 71/101, i.e. by only 0.3%. The change in cost allocation follows the change in capacity share.) The result is that the disincentive to invest in new South Island generation is reduced for generators that already have a significant South Island portfolio. A linear approximation is used to model this effect in GEM. Possible future plants have been assigned to generators for this purpose; where the likely owner of a project is not clear, it has been listed as 'other'.
Other transmission charges are not currently modelled in GEM.
Generation scenarios
This section describes the current versions of the five scenarios as at October 2007.
The schedules of generation builds are illustrated in the plots below.
The next plot shows how the average renewable energy percentage changes over time.
Sector greenhouse emissions vary more or less inversely, but are also affected by the coal/gas mix (i.e. they drop substantially if the coal-fired units at Huntly Power Station are decommissioned and not replaced):
The average annual electricity production is broken down by generation type in the plots below.
The model's build and retirement schedules
Background
In September 2006, the Commission developed draft grid planning assumptions (GPAs) as an initial step in preparing the 2007 Statement of Opportunities (SOO). This included the preparation and publication of demand forecasts and four generation scenarios for consultation with interested parties.
In May 2007, the Commission decided to delay the publication of the next SOO until the NZ Energy Strategy was finalised.
Now that the Government’s climate change policies have been developed further and the New Zealand Energy Strategy has been finalised, the Commission is recommencing work on preparing the next SOO, which is likely to be published in 2008. As an initial step, and following the receipt of additional information from Transpower, the Commission has revised the generation scenarios and other key inputs to the draft GPAs that were released in May 2007. This page contains those revisions.
The Commission and Transpower have also been discussing the draft GPAs published in May 2007 in the context of identifying scenarios to be used in applying the grid investment test (GIT) to a number of investment proposals, including an anticipated HVDC investment proposal.
In particular, the Commission is considering a "90% renewables by 2025" scenario which models the NZ Energy Strategy target of 90% of electrical energy produced from renewable resources by 2025. That scenario would be additional to the four scenarios anticipated by the draft GPAs. The weightings of the four draft GPAs released in May 2007 would therefore be affected.
The Commission is also considering revised assumptions regarding peak capacity requirements when modelling possible future generation developments in the scenarios anticipated by the draft GPAs released in May 2007.
More detailed work will be carried out over the next few months working towards a publication of the SOO in 2008, beginning with the release of an updated demand forecast.
Contact Brian Bull for any further information on the Commission's work programme on generation scenario development.
Generation scenarios - Outline
This section provides a high-level description of the five scenarios currently under development. More detailed descriptions of the scenarios are given at the bottom of the page.
|
Scenario |
Description |
|
90% Renewable by 2025 |
Government policies strongly discourage the development of fossil-fuel-based generation, and raise the proportion of renewable electricity generation to 90% by 2025. The dual-fuelled (coal-burning) units at Huntly Power Station are decommissioned between 2013 and 2019 and replaced by renewable generation. Geothermal, hydro and wind generation all feature strongly, with biomass-fired cogeneration, marine, and coal with carbon sequestration added later in the scenario. In the later part of the scenario, both renewable and thermal projects are added to provide peaking capacity in the North Island (including pumped and peaking hydro schemes). Demand-side measures also contribute to peak management. |
|
75% Renewable |
Government policies discourage the development of fossil-fuel-based generation. Combined with a constrained gas supply, this leads to the development of renewable options. Geothermal, hydro and wind generation all feature strongly, with the majority of the new hydro projects being located in the South Island, but all the geothermal and most of the wind in the North Island. In the later part of the scenario, both renewable and thermal projects are added to provide peaking capacity in the North Island (including pumped and peaking hydro schemes, and gas- or oil-fired thermal units). Demand-side measures also contribute to peak management. The coal-fired units at Huntly Power Station remain in operation until 2030. |
|
Mixed Technologies |
Government policies provide more moderate support to renewable generation. A mixture of generation technologies is the result, including new coal-fired generation in the North Island after 2020, as well as geothermal, wind, and hydro. Thermal peakers support intermittent generation. Demand-side measures also contribute to peak management. The coal-fired units at Huntly Power Station remain in operation until 2030. |
|
High Gas Discovery |
Timely and extensive exploration for gas leads to a relatively unrestricted supply of natural gas at prices similar to today's. Major new gas-fired power stations are constructed, with five new CCGTs installed by 2026 (including Huntly e3p and Otahuhu C). Government policies supporting renewable generation also encourage the development of geothermal resources, and, to a lesser extent, wind and hydro power. Demand-side measures contribute to peak management. The coal-fired units at Huntly Power Station remain in operation until 2030. |
|
SI Surplus Renewables |
This is a variant of the 75% Renewable scenario, and is similar in many respects (with a strong emphasis on geothermal, wind, and hydro generation). The key difference is that in the SI Surplus variant, the Tiwai Point aluminium smelter ceases operations with a gradual phase-out from 2014 to 2019. The results include increased northward power flows, and delays in generation build relative to the other scenarios. This scenario could now be considered to have lower probability than the other four, given the recent announcement that Meridian Energy and NZAS have signed a contract for delivery of power to the smelter until 2030. |
Generation expansion methodology
Modelling approach
In order to develop each scenario with a reasonable degree of internal consistency, the Commission has developed a Generation Expansion Model (GEM). GEM is a mixed-integer linear programming model and is used to find a least-cost schedule of the timing and location of new generation capacity, subject to energy and capacity constraints. GEM's data inputs include:
- lists of existing and potential future generation
- electricity demand forecasts
- limits on the performance of various generation technologies
- fuel prices and greenhouse costs
- projected production from existing hydroelectric generators under a range of inflow scenarios
- plant retirement dates where applicable, and
- various financial and economic assumptions.
The generation scenarios are cross-checked using SDDP (a commercial medium-term hydrothermal dispatch model; working SDDP input files), with a particular focus on unserved energy, the use of hydro storage and HVDC flows. They will also be checked using the Commission's in-development capacity model.
A notable feature of all the scenarios is the extent to which capacity is added to meet peak demand requirements rather than to meet overall energy requirements (resulting in, for example, the introduction of increased demand-side participation, thermal peaking generation and eventually pumped hydro). These outcomes reflect a potential transition of the New Zealand power system from being "energy constrained" to facing both energy and capacity constraints. Contributors to this change include the emphasis on base-load plant over the last 15 years and the amounts of new baseload and intermittent plant being added in the scenarios. The GEM model includes explicit capacity constraints, which have a major influence on the development of the scenarios. Previously these capacity constraints were designed to require that N-2 security be achievable at cold-year winter peak demand. However, the Modelling Group has reconfigured the GEM model to use a N-1 standard, which is considered to be a more reasonable assumption.
The Commission is aware of some comments to the effect that it is not clear that the current market design would lead to a level of generation investment that met the modelled capacity constraints. Other Commission work programmes are currently assessing this issue.
While the particular assumptions adopted in relation to a specific investment proposal will be considered by the Commission in the course of carrying out its decision-making responsibilities under part F, for the purposes of continuing the Modelling Group’s work the Commission will continue to assume that security of supply standards will be maintained by some means. Similarly, all scenarios will include enough generation to supply a 1-in-60 dry year (given a reasonable level of demand response).
Generating technologies
GEM models a wide range of large-scale electricity generation technologies. The main technologies included in the scenarios are:
- Thermal generation
- Coal (black coal in the North Island or lignite in the South)
- CCGT [combined cycle gas turbine]
- Peaker (provisionally assumed to be diesel-fired OCGTs, though in practice other technologies could be more cost-effective)
- Coal with carbon storage and sequestration (not before 2030)
- Cogeneration (gas- or biomass-fired)
- Geothermal
- Wind
- Hydro
- Backed by storage
- Run of river
- Modification to an existing hydro scheme to add peaking capacity
- Pumped hydro (adding firm capacity at peak, as well as shifting load from peak to off-peak periods and helping to balance intermittent generation)
- Wave
Demand-side measures are modelled as another 'generation technology'. Three main types of demand response are included in the scenarios:
- Dry-year demand reductions (priced at $500/MWh, with up to 500 MW available per island)
- Additional interruptible load (available to meet peak only, with another 100 MW potentially available by 2020 and more thereafter)
- Price-responsive demand reductions, which may include price-responsive industrial load, smart metering with advanced tariffs, and/or technology advances (available to meet peak only, with 150 MW potentially available in the North Island by 2020 and more thereafter).
The scenarios assume fuel prices of $4/GJ for black coal, $1.80/GJ for lignite, and $25/GJ for diesel. Gas prices are assumed to increase over time, in response to diminishing supply and (in some cases) exposure to international markets. The gas price rises from $5.5/GJ initially to $7/GJ in 2012 in all scenarios. It continues to rise to $10/GJ by 2020 with a limit of 75 PJ/year, except in the 'High Gas Discovery' scenario, where the price remains at $7/GJ with a limit of 120 PJ/year, and the 'Mixed Technologies' scenario, where the construction of an LNG terminal in 2020 allows 'unlimited' LNG imports at a price of $10.50/GJ.
For some plant (notably gas generators in or north of Auckland) fuel transport costs are included.
Carbon costs are additional to the fuel prices above. GEM models the cost of carbon relatively simplistically, as a fixed cost per tonne CO2-equivalent. This cost is introduced at a low level in 2010, rises to a maximum level in 2018 and remains constant thereafter. The maximum level is $15/t CO2-equiv for the Mixed Technologies and High Gas Discovery scenarios and $40/t CO2-equiv for all the Renewable scenarios.
The carbon costs assumed in GEM predate the Government's announcement of the Emissions Trading Scheme and may need review. However, since the actual cost of carbon under an Emissions Trading Scheme (ETS) would be determined by the market and could vary from year to year, it may not be possible to produce more accurate estimates at this stage.
These fuel and emission costs are also given in the GEM data spreadsheet, which is distributed with the GEM model.
Modelled generation projects
The scenarios are built based on lists of generating plants. These fall into into three categories:
- 'Existing' plant that were already in operation at the start of 2007
- Plant that have been commissioned since the start of 2007, are committed, or are highly likely to be built in the next few years
- Plant that may be constructed in future.
- Huntly e3p (385 MW gas turbine, commissioned in 2007);
- Mokai 3, Ngawha 2 and Kawerau (120 MW of new geothermal capacity by 2009);
- Potentially a Rotokawa expansion, Te Mihi replacing Wairakei, Tukairangi Rd, and/or Tauhara (a further increase in geothermal capacity of as much as 250 MW by 2011);
- Deep Stream and probably the Hawea Control Gates project (22 MW hydro by 2010);
- White Hill, Tararua 3, West Wind, and Te Rere Hau (as much as 340 MW by 2009).
A key input to this list of projects was the following report by PBA, providing information on existing and potential thermal, geothermal, wind, and hydro schemes.
Readers should note that the inclusion of a project on the list does not necessarily mean that any developer is considering that project. In fact, some projects on the list have been specifically disavowed by generators. Nonetheless, they are included for completeness and in case they are taken up by developers in the future.
The complete list of possible future plants in the model is included in the GEM data spreadsheet, which is distributed with the GEM model. The spreadsheet includes assumptions on project capacity, cost structure, resource quality, and earliest possible commissioning date.
Many of these generation assumptions have changed since the publication of draft GPAs in May 2007. They are potentially contentious, and the Commission welcomes feedback or further information.
At time of writing, the list of projects includes:
- eight sites for standard coal-fired plants (all but one in the North Island) and two for lignite-fired plants (both in the South Island);
- five sites for coal plants with carbon storage and sequestration (not before 2030);
- seven sites for combined cycle gas turbines (some replacing existing plant at the end of their lives);
- diesel-fired OCGT peakers, in blocks of 150 MW;
- two sites for gas-fired cogeneration, five for biomass-fired cogeneration, and two for other forms of cogeneration (altogether adding to 360 MW);
- various geothermal projects (with as much as 1100 MW of new capacity by 2030 in some scenarios);
- enhancements to existing hydro schemes in the Waikato, Clutha, Waitaki and Manapouri systems;
- over 50 possible new hydro schemes (some considerably more economic and/or consentable than others);
- three 300-MW pumped hydro schemes at nominal locations in the central North Island;
- several dozen wind projects (totalling to nearly 4000 MW, though some are more economic than others, and no scenario includes all or even most of them);
- demand-side options; and
- as much as 300 MW of wave generation.
Note that GEM does not model relatively small embedded plant (such as backup generators, very small hydro schemes, and distributed wind generation). The electricity demand forecasts implicitly assume that the contribution from embedded plant is maintained at current proportions, which would require the gradual addition of distributed generation.
Retirements
GEM does not model plant retirements endogenously; these must be provided to the model as input assumptions. Currently, all new projects and most existing stations are assumed to remain in operation until 2042 at least. The exceptions are;
- New Plymouth Power Station is assumed to decommission in 2020;
- The four dual-fuelled units at Huntly Power Station are assumed to decommission in 2030;
- Taranaki CC and Otahuhu B are assumed to decommission in the 2030s (perhaps to be replaced by new CCGTs on the same sites).
- Wairakei is assumed to decommission in 2011 (potentially replaced by Te Mihi).
We acknowledge that all these retirement assumptions are highly speculative. In fact, the owners of these plants have not stated when they plan to retire them.
Energy constraints
GEM requires that scenarios include enough generation to supply demand in a dry year, given modelled levels of demand-side response. Further generation can be added on an economic basis and/or to satisfy the capacity constraints.
The build schedules in the scenarios are optimised using a two-step process. Firstly, project build times are selected on the basis of optimality (minimum discounted post-tax cost) in a model run where hydro inflows are assumed to be 97% of the historical average. Secondly, thermal peakers can be introduced or brought forward on the basis of optimality in a model run with dry-year (1932) hydroelectric production. The intention of this process is to model a future in which baseload is constructed to meet normal levels of need, but additional thermal peakers can be constructed for dry-year reserve (or to meet unexpectedly high levels of demand growth).
(Dispatch simulations are then carried out using a range of hydro inflow scenarios, as discussed in the GEM documentation.)
GEM models electricity demand via load duration curves. Currently these load duration curves are quarterly, with four load blocks per quarter (though GEM will soon allow more flexibility here). The total annual demand in each island is based on the draft GPA demand forecast, with allowances for AC losses and to avoid double-counting of modelled embedded generation. The shape of the load duration curve is based on a historical reference curve.
The demand forecast is included in the GEM data spreadsheet, which is distributed with the GEM model.
Electricity production from existing hydro plant is estimated outside the GEM model. The SDDP model was used to estimate monthly electricity output from the existing hydro system, under a range of historical inflow sequences from 1932 to 2004. These production figures have been pooled for each island and entered into GEM as 'scheduled hydro'. The assumption is that existing hydro plant will continue to operate much as it has in the past. Hydro storage from one quarter to another is not explicitly modelled in GEM (production cannot be carried over to a subsequent period) – however, the interperiod storage dynamics were modelled in SDDP and are incorporated in the production figures.
Capacity constraints
GEM includes three capacity constraints; these are 'hard' constraints which must be satisfied by all build schedules. They require a 'N-1 at cold-year winter peak' standard – in other words, generation must have the technical capability to serve the maximum likely winter peak demand in a cold year, with enough margin to cover a single contingency. The three constraints each model a different contingency:
- there must be enough capacity nationwide to cover the loss of the largest generating unit;
- there must be enough North Island capacity to cover the loss of the largest North Island generating unit; and
- there must be enough North Island capacity to cover the loss of one HVDC pole.
These constraints affect the build schedule in several ways. They bring in demand-side options; later in the scenarios they lead to the construction of thermal peakers and/or pumped and peaking hydro; and they add an incentive to favor mid-order plant over baseload. All these effects apply more strongly in scenarios with significant amounts of intermittent generation (wind, wave and/or run-of-river hydro), and they would also apply more strongly in scenarios with less HVDC capacity.
The peak demand figures used in the forecasts are derived from the annual half-hourly peak demand forecasts published as part of the draft GPAs from May 2007. The GEM peak forecast was produced by starting from the Commission's prudent forecast from 2007 and growing it forward at the growth rate of the expected forecast. This was intended to predict cold-year demands under a scenario of average underlying demand growth (as opposed to using the prudent forecast throughout, which would model a scenario of high underlying demand growth). The half-hourly forecast was then increased by 90 MW (national) or 60 MW (North Island) to allow for within-half-hour variation. An allowance for AC losses was added on, assuming AC transmission losses at peak of 4.0% in 2007, trending upwards to 5.5% in 2030 due to increasing system loads.
A peak contribution factor has been specified for each technology, for use in these constraints.
- Thermal stations (other than cogeneration) have been assigned a peak contribution factor of 0.95. This is intended to reflect the possibility that these stations may be subject to forced or planned outages, or for some other reason may not be available at full capacity (in some or all units) during system peaks.
- New cogeneration has been assigned a peak contribution factor of 0.6, reflecting the link to the associated industrial process and the possibility that it will be unavailable, or at least not operating at maximum output, during peak periods.
- Geothermal has been assigned a peak contribution factor of 0.9, reflecting typical levels of availability of existing geothermal plant during peak periods.
- Existing hydro 'HydSC' has been assigned a peak contribution factor of 0.95. Actually this figure was somewhat arbitrary; the 'nameplate MW' and peak contribution factor were chosen so that their product should equal the estimated peak contribution from existing hydro assets. This estimated peak contribution is based on historical availability during peak periods (including availability in the reserve market).
- New pumped and peaking hydro has been assigned a peak contribution factor of 100%. Other new hydro backed by significant storage (such as the various control gates projects or the North Bank Tunnel) was assigned 0.95. All other new hydro was classified 'run of river' ('HydRR') and assigned a rather lower peak contribution of 0.65, reflecting the possibility that water will not be available at peak times. This figure of 0.65 was arrived at by exploratory analysis of some existing run-of-river plant (the Commission constructed a historical availability curve and convolved it with an approximate availability curve for the rest of the generation stack to derive the incremental effect on total generation availability).
- Wind was assigned a peak contribution factor of 20%. There is a lively debate about the impact of increased wind penetration on the New Zealand power system, and (recognising the limited extent to which the effect of more intermittent generation on system security can be reduced to a single number!) on the appropriate peak contribution factor for wind. Some parties have suggested that the correct peak contribution of wind in the New Zealand power system may be as high as 40% for small amounts of additional wind; others, that the correct figure is zero. We are currently doing more work on this issue (considering the possibility that wind may be operating at low output during peak times, the probabilistic impact of this on the total generation supply stack, and other impacts of wind generation such as potential increased requirements for frequency keeping and/or balancing) and consider that the 20% figure is an appropriate number to use in the interim.
- we have moved on from a more simplistic formulation using 'incremental capacity constraints' to the new constraints which identify specific risks;
- the demand estimates and peak contribution factors have been revised; and
- the Commission has moved towards the view that 'N-1 capability at cold-year peak' is an appropriate standard for generation scenarios, rather than 'N-2' as in the May 2007 draft GPAs. The N-2 standard was rejected because we consider that there is yet insufficient evidence to conclude that such a high level of redundant capacity would be required to provide peak supply security.
In all scenarios, carbon charges have the effect of encouraging the production of electricity from renewable sources.
The '90% Renewables by 2025' scenario also models an explicit requirement for 90% renewable electricity by 2025. The required renewable percentage increases linearly from 70% in 2010 to 90% in 2025 and remains at 90% thereafter. This is modelled as an average-year requirement, not a dry-year requirement – the renewable percentage is allowed to dip below 90% in dry years, but is over 90% in wet years and averages out to approximately 90% in the long run.
For this purpose, 'renewable' technologies are deemed to include geothermal, wind, hydro, marine, and coal with carbon storage and sequestration. Other forms of thermal generation are still used for peaking and balancing, but do not count as 'renewable'.
WACC and discount rate
The generator WACC assumed in developing the generation scenarios is 8% real post-tax. This is slightly higher than the earlier assumption of 7.7%, to notionally account for hurdle rate issues as well as to make it distinctly different from the 7% assumed for transmission investments.
Concept Consulting has prepared advice for the Commission on this issue.
Note that GEM can be instructed to process results using different discount rates such as 4%, 7% or 10%, but these should not be confused with the 8% generator WACC used in the objective function.
Locational effects
Location factors have been used to reflect the effects of locational price on project economics. Each modelled project is mapped to a transmission region, and each region has a location factor associated with it. For example, an identical project would require a lower average price to be constructed were it located at Otahuhu than it would were it located at Tiwai. The location factors used are constant and do not change in response to generation and demand changes.
Note that cost outputs can be produced exclusive of these location factors in the current version of GEM (unlike earlier versions).
AC transmission and distribution losses are built into the demand forecasts used in GEM, using island-level loss figures (rather than being based on the location factors above). Losses on the HVDC link are modelled explicitly using a piecewise linear approximation to the loss function.
Capital cost uncertainties
There is considerable uncertainty around the capital costs and potential benefits of the modelled wind and hydro generation projects, which is not reflected in the baseline project cost/benefit estimates in the list of modelled projects. As a consequence, the unmodified cost curves for wind (and hydro, to a lesser extent) are very flat, with relatively little variation in LRMC between projects. In reality these curves should be more sloping, with some potential projects delivering more or less benefits than shown, or able to be constructed at a lower or higher capital cost.
In the scenario development, this is implemented as follows. Modelled wind and hydro projects (excluding committed and 'highly likely') are divided into mutually exclusive subsets of roughly equal size, on an arbitrary basis. In each scenario, one of these subsets is selected, and it is assumed that projects in this subset possess unexpected advantages (e.g. can be constructed more cheaply, or can produce more energy, than the baseline predictions indicate). These advantages are modelled by reducing the capital costs of the selected projects by 10% for hydro, or 20% for wind (where the uncertainties are considered to be greater). The effect is to shift the selected projects higher up the merit order. This has the effect of increasing the slope of the cost curves for wind and hydro, with the affected projects becoming more competitive with other forms of generation (i.e. thermals). A useful secondary effect is to increase the diversity between scenarios - each scenario has a different set of projects with reduced costs, and hence a different build order.
Transmission charges
South Island generation projects are assumed to pay a charge of up to $40/kW p.a. for the HVDC. The figure of $40/kW was derived by annualising estimates of Transpower's future HVDC charges and apportioning them to an assumed South Island installed capacity. (This figure will be reviewed soon in light of indicative cost recovery figures provided by Transpower.)
For generators who already have significant installed capacity in the South Island, the marginal increase in HVDC cost allocation per MW of new generation is less than $40/kW. (Consider a new plant with capacity equal to 1% of installed South Island capacity. If this plant belongs to a new entrant generator, then their share of total capacity rises from 0/100 to 1/101, i.e. to almost 1%. If, on the other hand, it belongs to a generator that already has 70% of island capacity, then their share of total capacity rises from 70/100 to 71/101, i.e. by only 0.3%. The change in cost allocation follows the change in capacity share.) The result is that the disincentive to invest in new South Island generation is reduced for generators that already have a significant South Island portfolio. A linear approximation is used to model this effect in GEM. Possible future plants have been assigned to generators for this purpose; where the likely owner of a project is not clear, it has been listed as 'other'.
Other transmission charges are not currently modelled in GEM.
Generation scenarios
This section describes the current versions of the five scenarios as at October 2007.
The schedules of generation builds are illustrated in the plots below.
The next plot shows how the average renewable energy percentage changes over time.
Sector greenhouse emissions vary more or less inversely, but are also affected by the coal/gas mix (i.e. they drop substantially if the coal-fired units at Huntly Power Station are decommissioned and not replaced):
The average annual electricity production is broken down by generation type in the plots below.
The model's build and retirement schedules
Last update on
06 November 2007 09:29 AM
+64 4 460 8860
+64 4 460 8879