Four indicators of "shortage" in relation to the modelling of the two scenarios were examined:
- SI Storage;
- Frequency of reserve generation dispatch;
- Demand response indicators;
- Capacity margin.
7.1 SI Storage Results
Under both demand scenarios, as shown below Table 10, dry year security of supply was maintained such that SI storage did not fall below 300GWh at all with 2.0% pa demand growth and at most once with 2.5% pa growth. However, SI storage did fall below 500GWh in up to 3 years out of 72 inflow sequences.
Table 10: SI Storage Minimum Levels
| Year | 2% pa Demand Growth | 2.5% pa Demand Growth |
|---|
| SI Storage < 300GWh | SI Storage < 500GWh | SI Storage < 300GWh | SI Storage < 500GWh |
|---|
| 2004/2005 | 0 | 2 | 1 | 3 |
| 2005/2006 | 0 | 1 | 0 | 3 |
| 2006/2007 | 0 | 2 | 1 | 2 |
| 2007/2008 | 0 | 3 | 1 | 3 |
Therefore, if the EC were to use 300GWhSI storage as an indicator of 1-in-60 security, then the security standard would be achieved, and in fact bettered, given our modelling assumptions. On the other hand, using a 500GWh test then the required security level would be achieved in only 1 of the four years with 2.0% pa growth, and none of the four years with 2.5% pa growth. However, it is worthwhile reiterating one of the key changed assumptions in this security update is that there is no demand response or reduction (not even a reduction by larger electricity consumers that may have some exposure to the high spot prices expected in a dry year).
It is also worth reiterating that all of the results below were produced with SI starting storage of 2,100GWh which is less than SI storage is likely to be on 1 April in most years.
The following charts show the forecast SI storage trajectories for the year 2007/2008. The storage trajectory that bottoms out late in September in both charts is the 1932 inflow sequence. The two that bottom out at similar or lower storage a month or more later are 1976 and 1937, the former being the lower of the two.
Figure 8: Forecast SI Storage Trajectories 2007/2008 with 2.0% pa Demand Growth
Figure 9: Forecast SI Storage Trajectories 2007/2008 with 2.5% pa Demand Growth
The charts above also show that there are a small number of inflow years when SI storage does not recover back to the target minimum of 2,100GWh33 notwithstanding thermal offering aimed at achieving this target. This is an area that the EC may need to investigate in some detail exactly what the target should be for end of year storage in the SI, and how it will be achieved?
7.2 Reserve Generation Results
Whirinaki operated in only a few years in line with the government's recent proposals for reserve generation.
Table 11 below shows the frequency of dispatch at Whirinaki under 72 inflow sequences in each demand scenario.
Table 11: Whirinaki Runs
| Year | 2% pa Demand Growth | 2.5% pa Demand Growth |
|---|
| 2004/2005 | 2 | 4 |
| 2005/2006 | 2 | 4 |
| 2006/2007 | 2 | 3 |
| 2007/2008 | 1 | 2 |
7.3 Demand Response Signalled as Non Supply Generation
Dummy generators are used in the modelling to represent "non-supply" which can also be taken as an indicator that some demand response is required. Two non supply generators are modelled, one in the NI and one in the SI. Both are offered at $10,000/MWh.
Non-supply generation did not get dispatched at all in either of the scenarios indicating that there would be no demand response required, even in the drier inflow years. This is not to say, however, that there would be no demand response in reality. When electricity spot prices climb in a dry year there may well be reduction in demand by organisations that are exposed to these prices.
7.4 Capacity Margin
In a market dominated by thermal generators with no constraints on output, one could monitor and enforce the security of supply standard by use of a capacity margin. The capacity margin can be expressed as follows:
A shortage would obviously occur when the capacity margin is less than zero.
In New Zealand, defining and applying a capacity margin would be difficult. The chart below shows the total modelled capacity of all the thermal, small and new generators for the 2004/2005 year. In addition, it includes the total installed capacity of the major hydro generators. The black line is the average monthly forecast demand and the blue line is the forecast of monthly peak demand.34 Losses have been added to the demand for illustrative purposes.
Figure 10: Total Installed Capacity Versus Demand
The capacity margin on the above chart, the gap between peak demand and total installed capacity, is large. However, it is very misleading given that the potential generation from hydro generators is highly dependent on inflows, i.e. we cannot expect hydro generators to run at full capacity for 100% of the time. In fact the output of any hydro generator is highly dependent on the current state of storage and potential inflows, which are inherently difficult to predict.
Another approach to examining security of supply is to look at the aggregate energy available and compare this with demand as opposed to capacity. The following chart shows the total potential energy from a dry inflow sequence plus all other available energy, i.e. assumes all stations are run at available capacity.35 The chart shows the demand as total energy rather than as peak demand.
Figure 11: Potential Energy Supply Versus Demand under the 2001 Inflow Sequence
The above chart is interesting in as much as it indicates the potential for a brief supply problem at the end of June 2004. On face value, one might think that hydro storage could more than adequately cover this temporary shortage, also providing that little bit extra during daily demand peaks. However, it should be noted that the above supply and demand is taken on an aggregate national basis, i.e. it does not reflect any supply problems that can be caused by regional constraints in the transmission Grid.
The chart below shows the potential output from all SI generators under the same "dry" inflow sequence (2001 year) charted against SI demand. The analysis also assumes full southward capacity transfers on the HVDC Link of 626MW at the sending end.
Figure 12: SI Demand Versus Available Energy under the 2001 Inflow Sequence
Given the limitations on southward transfers on the HVDC Link, it becomes evident that SI demand would have to be met by SI storage in the regions where the capacity margin is shown as being "in the red." In this example, a target capacity margin could be calculated where SI demand exceeds the deliverable energy from inflows.
Simplistically, dry year security of supply could be accommodated by ensuring at least the capacity margin is maintained as storage in the SI lakes. This approach, however, would be overly simplistic:
- The analysis assumes that all stations, including the reserve generation stations would be required to run at available capacity36 all of the time;
- If the capacity margin is calculated against the driest year in 60, and storage limits enforced, there would be significant spill in other inflow years;
- "Worst Case" inflow sequence analysis assumes that the "worst case" is represented by historical data;
- Whilst this analysis included the transmission limitations of the HVDC Link, there are other lines which cause regional effects from time to time, even within the SI.
- It does not take into account random plant and line outages which, depending on their length, timing and severity, could significantly disrupt the capacity margins.
The conclusion is that the capacity margin over the nation as a whole is only just sufficient for a dry year if nothing else goes wrong, and it is insufficient in the SI due to constraints on the Grid.
Thus the management of SI hydro storage lakes is critical. In our modelling we assumed that key SI lakes would operate on the assumption that somewhere between 300GWh and 500GWh should be left in the SI lakes even in the driest of years, if no other adverse events occurred.
7.5 Transmission Constraints
Examination of the modelled results of inflow years such as 1932 shows that power must be transferred from the NI to the SI at high levels for sustained periods. The limits on the HVDC and on the total transfer of power from BPE to HAY place significant constraints on these transfers and potentially limit the ability of reserve plant such as Whirinaki in its designated role as reserve generation.
7.6 Thermal Response
A second issue shown up by the modelling is that plant such as Huntly and New Plymouth will need to run hard and early. Figure 13 below shows the average weekly output that the modelling has produced for these stations for 2004/2005 with 2.0% pa demand growth.
Figure 13: Huntly and New Plymouth Response with 1932 Inflows
Huntly is modelled with lower offers than New Plymouth and it ends up running at full output of 1,000MW from June through to end of September when a 250MW unit is taken out for the first of a number of planned outages that take place each year from October through to end of March.37
While this could be difficult to achieve in reality, note that there is spare capacity in New Plymouth which is modelled as offering 400MW outside of planned outages. Using average weekly output, however, as shown in the chart, masks the fact that any problems with these stations would have a greater impact during daily peaks than is indicated. This is where plant of last resort such as Whirinaki (and also including the Otahuhu A station) becomes valuable even if it does not show up as running often in our modelling.
A related consideration is that spinning reserve is normally provided by plant that is not running at 100% rated output. It would be impossible for Huntly, for example, in the scenario shown in Figure 13, to provide spinning reserve for almost 5 months.
The requirements for reserve generation be it spinning reserve or frequency keeping reserve has not been explicitly modelled and as such, it should be noted that the modelling may overstate the available capacity of plant and hence understate the risks. This is discussed in more detail in section 5.6.