• Home
• Energy & Resources
• Electricity
• Archive
• Reports
• Dry Year Security Reports
• 2004/2005-2007/2008
• 5. Modelling Assumptions

• Electricity

• Archive

• Reports

• Dry Year Security Reports

This section details the general assumptions that have been made in the market modelling. All assumptions outlined in this section, unless otherwise stated, pertain to the base line scenario which represents "reasonable" market behaviour.

Key variations to the modelling in our variation scenarios relate to the rate of demand growth over the next four years, the amount of demand response and whether new generation assets are built between now and 2007.

Unless otherwise stated, all assumptions detailed in this report have been derived from ELL's independent assessment and view of the electricity industry and market and discussed with the MED.

5.1 Existing Generating Stations

All generating stations that are connected to TPNZ's Grid have been modelled - we do not model generators which are embedded into local networks because raw demand data is taken from the NRM which is net of embedded generation - refer section 5.3.

Table 4: Modelled Generating Stations

The smaller generating stations have been modelled as offering their generation at base load prices which ensures dispatch. This strategy is consistent with our observation of smaller generators in the market since October 1996. Offer quantities for the smaller generators have been derived from average output since 1996.

5.2 New Generation Assets

The following new generation assets have been modelled in our scenarios.

Table 5: New Generation Assets

The single largest new plant in the above schedule is Genesis' e3p plant. As with other new plant, there is uncertainty as to exactly if and when the plant will proceed. This has implications for security risks for 2006/2007 and 2007/2008.

5.3 Generator Offers

Generator offers are a key input to the modelling process given that it is the marginal generator's offer price¹⁸ which primarily determines the nodal price from one period to the next. Offer prices also determine the order in which plant is dispatched - the plant with lowest offer is dispatched first, followed by plant with progressively higher offers.

The first security update in December 2002 was implicitly based on the assumption that the market would deliver security of supply. In that context, it was essential to model realistic market offers, however, things have changed and security of supply in the future will be delivered by the EC working with the industry. In this context, and given that the purpose of the modelling was to estimate the physical risks surrounding dry years, we have, in this update, focussed on establishing the order in which plant would be dispatched based on SRMC, and ignored any attempt to make offers that reflect the price outcomes in the market.

With the exception of the generators listed in Table 6, the major hydro generators and smaller generators for which we have inflow data, all generators are assumed to offer a profile or constant offer at $0/MWh.

Offers for the main thermal plant reflect a crude price order based on SRMC, estimates for which are made from publicly available information. For large, multi-unit stations such as Huntly and New Plymouth, the range of offers is increased to provide separation of prices for convenient diagnosis of problems with modelling runs. Note that Whirinaki is only offered from 1 June 2004.

Table 6: Main Thermal Generators and Their Modelled Offers

5.4 Generator Outages

Generators are taken out of service a number of times each year for routine, planned maintenance, known as planned outages. They also suffer a varying number of unplanned, unexpected outages each year. Usually, though not always, the total duration of the planned outages in a year far exceeds the total duration of unplanned outages. In the previous security update published in December 2002 the modelled capacity of each large generator in Table 6 was adjusted down to account for outages. In this update, however, to account for the effect of timing of long outages, the planned outages are modelled explicitly and the rest of the year the plant is assumed to be 100% available.

5.5 Generator Availability

Effectively we have assumed that all of the plant in Table 6 will be available to run whenever it is required. Indeed, since the supply-demand balance is relatively tight, plant such as Huntly and New Plymouth will need to be run early and hard to maintain security in the more severe dry years such as occurred in 1932. Also, given that there could, on occasion, be major plant failures, it is important to note that the results presented in this report will tend to understate the effect of adverse events in future, which is why the indicators for shortage include SI storage of 300GWh and 500GWh, leaving some room for such events.

5.6 Other Types of Reserve Generation

Other types of reserve, such as frequency keeping reserve, spinning reserve, tail water depressed reserve and interruptible load reserve are currently required to cover the risk sudden generator outages and/ or one pole of the HVDC Link failing. This reserve capacity is calculated¹⁹ as part of the optimum dispatch of generation to meet demand in real time.

The amount of reserve required is dependent upon a number of other things such as the contingent risk in each island and system inertia.

In simple terms, the largest single contingent risk in each island is calculated either as:

• the single largest generating unit, usually OTAB in the NI and Manapouri in the SI; or
• one pole of the HVDC Link failing when transfers exceed a prescribed limit.²⁰

In summer night periods when there is a relatively low level of generation, the effect of one large generator tripping is potentially more severe than during a winter day period, when one would expect many generating units to be dispatched to meet high demand. Therefore the amount of reserve required in the low demand periods is usually greater in low demand periods than that required in the periods of high demand.

Frequency keeping reserve generation is also required and is directly contracted between generators and TPNZ. It effectively requires 50MW of capacity to be kept spare in each island. Contracted generating companies offering this type of reserve are required to keep 50MW of capacity spare in case they need to increase or decrease output to maintain frequency within prescribed limits. It should be noted that the capacity can be reserved at more than one station at a time, as long as the combined total is equal to 50MW.

The requirement for all four types of reserve generation discussed here has not been explicitly modelled in the scenarios presented in this report. The significance of this exclusion is difficult to assess without explicit modelling but it should be noted that it could affect the results, in as much as the modelling may overstate the available capacity, especially in those periods where supply is tight.

5.7 Fuel Supplies

Earlier this year fuel supplies were in doubt but now it appears that all of the plant in Table 6 is able to run at full output as much as required, at least through 2004. Huntly is able to run at full output on coal and New Plymouth on oil, with gas used in this plant when it is available.

For later years it has been assumed that there is enough thermal fuel for thermal plant to be unconstrained by fuel availability. This includes there being enough fuel (gas or possibly distillate) for Huntly e3p (from 1 April 2004) and the two existing combined cycle stations, OTAB and TCC. Should there be thermal fuel constraints, then clearly risks will rise above those indicated in this report.

5.8 Demand

Historical demand from the NRM for 2002/2003 was used as the base figure from which to derive demand for 2004/2005 through 2007/2008. The NRM metered demand that is used is net of embedded generation,²¹ so our forecast figures may appear slightly less than demand figures published elsewhere²² which may apply to gross consumption.

The following table shows the NRM's total demand and demand growth for the last 6 April/March years. The 2002/2003 figure can be extrapolated from the 1997/1998 assuming compound annual growth of 1.94%.

Table 7: NRM Annual Demand

Demand reduced this year in the period April through June because of the call for savings, but in July demand was up 2.0% on July 2002 indicating that demand growth is back to around 2%.

On the other hand, if a line is fitted to the demand data above then it has a slope equivalent to 2.3% pa growth. Accordingly, a lower rate of 2.0% and an upper rate of 2.5% were chosen for the two scenarios in order to cover likely extremes.

Total demand is allocated to EMarket's 184 nodes using the half hourly demand profile from 2002/2003 which does not include any periods when widespread savings were called for.

Table 8: Modelled New Zealand Demand

Note that 2.0% demand growth is equivalent to about 87MW additional base load capacity each year (excluding embedded demand), and 2.5% is equivalent to about 111MW each year.

5.9 Demand Response

Demand for electricity is very inelastic. That is to say that demand response to spot price is not significant until prices rise very high. In past recent "crises" in 1992, 2001 and this year, the larger consumers exposed to spot prices have voluntarily reduced demand and the government has called for widespread voluntary savings from consumers shielded from rising spot prices.

In contrast to the first security update in December 2002, in which we modelled demand response occurring at prescribed wholesale prices, in this report we have taken a conservative approach and removed all demand response, even though this may not be entirely realistic.

It may be noted that ELL's understanding is that the recent announcements about dry year reserve generation initiatives are aimed at ensuring that demand reductions (apart from those made by larger consumers in response to high spot prices) will not be necessary unless a dry year event is worse than, or likely to be worse than, a 1-in-60 event.

5.10 The Grid

EMarket models a 184 node composite grid connected by over 249 lines as shown in the maps and tables in Appendix A: Grid Nodes and Lines. In the December 2002 report, a Grid with only 38 nodes was used but the larger Grid was used this time to allow flows on lines that could constrain more often in dry years to be examined. The actual grid has around 488 nodes joined by over 800 lines and transformers. To model the actual grid to this level of detail would take considerably more time and produce little benefit in terms of increased accuracy in the resulting security assessment.

Figure 3: BPE-HAY Area

By using a composite grid, we are able to model various scenarios over many inflow years with acceptable accuracy.

Only the 16 lines shown in Table 9 were set to constrain at their limits and these were chosen based on their impact on dry year security and also on the ability of thermal plant in the NI, including Whirinaki, to export power to the SI.

Not all of the line limits shown are the limits of the individual lines. For example, the lines from Bunnythorpe (BPE, near Palmerston North) to Haywards (HAY, just north of Wellington) at the northern end of the HVDC link, are governed not by their individual limits but by a series of limits on the total power flow from BPE to HAY on four main lines running at 220kV. These lines are shown in Figure 3 running from BPE direct to HAY (two lines modelled as one) and from BPE to LTN and then down to HAY direct (one line) and to HAY via WIL (one line.) The limits shown in are set on the modelled lines BPE-HAY and BPE-LTN in order to achieve the same result in EMarket as the operational constraint on the total flow on the four lines.²³

Table 9: Lines with Constraints Enabled²⁴

Key:    SN Summer night; SD Summer day; WN Winter night; WD Winter day

Transpower's winter period runs from around 1 May through to 30 September.

The fit on these lines relative to the actual limit on the total is very good so that the operation of the Grid from BPE right down to Benmore (BEN), including the HVDC link running from HAY to BEN, is modelled accurately. This is critical because the capacity of the Grid from BPE to BEN is a major limiting factor on the amount of NI thermal generation that can be sent south in a dry year.

Note that the overnight limit on total transfer from BPE down to HAY, and the 626MW limits on south transfers on the HVDC, have not been made operational as yet but we assume in the modelling that it will be, as planned, once operational protocols are established. Even if this turns out to be incorrect, we still believe it reasonable to assume that it would be made operational in the event of an extreme dry year event and imminent shortages.

The same method applies to the limits on CML-CYD (Cromwell to Clyde - shown in Figure 4) but because of the large amount of generation around this area of the SI the approximation is not as accurate as it is for BPE to HAY. However, the impact on security of supply of these SI lines on the overall result is not as significant.²⁵

Figure 4: CML-CYD Area

5.10.1 HVDC Capacity

From 12 November this year Transpower will make operational a scheme that will allow the HVDC link to carry more power south than it could in December 2002 when the first security update was published. This, plus other operational changes made, will allow the HVDC to carry up to 626MW when Wellington load is low, which generally occurs overnight. At other times the limit will be lower but in our modelling the HVDC link only runs at or near 626MW²⁶ south over night anyway.

5.11 Water Values

Probably the most striking feature of this country's electricity supply system is that over 60% of the total generation is by hydro electric stations that can store only about 12% of the total inflows for the year on average. This makes the management of these reservoirs critical, whether or not operating in a market environment.

A hydro generator with seasonal storage has virtually no real marginal costs, but has to manage a finite "fuel" resource, generally limited to expected inflows when planning ahead, albeit with the potential for a wide variation.

The technique usually employed is to calculate the opportunity cost of the water in the hydro system's reservoirs at any point in time, also known as the "water value." Hence, the water value is the expected future value of the water in the reservoir. In principle, if the spot price is greater than or equal to the water value then water should be released for generation, otherwise it should be held in storage.

The algorithms employed by our EMarket model used in this modelling are based on these water value optimisation techniques but taking into account the volatility of inflows, thermal and other hydro offers, demand and dry year security of supply.

The four major hydro systems'²⁷ offers are based on the opportunity cost of water. The aim of the water value calculation in EMarket is to determine the position of marginal water value contours that correspond to the prices of other major offers on the grid.

The calculation of water values is a two-fold process. Initially, water value contours are calculated based upon all available generation, all available historical weekly inflows,²⁸ the security of supply parameter specific to each system and corresponding thermal offers in the market.

The security of supply parameters used in the water value optimisation are specific to each hydro system and are effectively another set of assumptions in the modelling. The security of supply parameters have been tuned to ensure realistic dispatch of hydro generation in relation to available thermal generation and to ensure "best" use of the water. They have been based on observation of past hydro behaviour. They do not incorporate anything new to provide for the recently announced 1-in-60 security standard.

The chart below shows an example of a set of water value contours from our EMarket model. Each contour has a fixed value which is the same as the thermal offer band to which it refers and can also be viewed as an "operating guide line" in the sense that, if thermal offers do not change between optimisation and operation, offering at the water value will ensure the optimal water release is achieved.

In Figure 5 below, if storage at Waitaki is around 1,200GWh at 24 June, the value of that water would be priced at between the New Plymouth offer price of 9c/kWh and the New Plymouth offer price of 11c/kWh. A linear interpolation between these two contours results in a water value, and offer price of 10c/kWh.

Figure 5: Water Value Contour Example