In a highly competitive generation market, one might expect that hydro generators with significant storage would be tempted to use more water in the winter, when prices are usually highest, than they currently do.
The evidence from our observations of the electricity market, suggests that in respect of Waitaki and Manapouri:
- storage is managed to ensure that sufficient storage is available to make up any shortfall in inflows, i.e. water is held back early in winter;
- coming into winter rising water values help ensure enough water is held back.
6.1 The Importance of Managing Storage
It is extremely important that the SI hydro reservoirs, in particular, are managed in such a way that there is adequate storage to prevent supply shortages during a period of low inflows. Storage makes up one half of the hydro equation with inflows being the other.
The following chart shows historical weekly inflow sequences from 1931 to 2002 which have been converted29 to total productive capacity by week in GWh. Maximum transfers south of 626MW, or 101GWh per week after losses were added to each weekly inflow. The adjusted inflows have been charted against actual forecast SI demand for 2004/2005 under the 2.0% pa demand growth assumption.
Figure 6: SI Inflows, HVDC Transfers South and SI Demand Including Losses
The red curve is the 2.0% scenario forecast SI demand. The other curves are weekly inflows to the major SI hydro systems plus continuous 626MW south on the HVDC link. Ignoring SI storage, where the curves are below the red line indicate that there is not enough energy available to meet SI demand.
The chart shows that SI demand could be met by SI inflows and south transfer on the HVDC link, without the benefit of SI storage, for no more than 15 weeks in a year.30 In the worst week in the depths of winter there are 40 historical weekly inflow sequences that are insufficient to meet SI demand. SI storage is also needed with continuous maximum HVDC south transfers.
In conclusion, this simple analysis shows that the management of hydro storage in New Zealand is of prime importance in maintaining secure electricity supply.
6.2 The Criteria for Shortage
The EC's target of 1-in-60 dry year security means, in simple terms, that there should only be a "shortage" 1 year out of every 60 on average in the long term. In order to assess the outlook for dry year security against this target, one should have an objective criteria by which to assess it. This is difficult to define for New Zealand for a number of reasons.
In an electricity supply system composed entirely of well maintained thermal stations, with access to more fuel than needed at a reasonable price, it is a relatively straightforward matter to total the available thermal capacity and then subtract off the expected demand for the coming year to get the expected "capacity margin." The criteria for shortage is stated simply as:
The task in assessing the outlook for security of supply ahead of time is one of predicting the capacity margin. As a refinement, one could allow for the possibility of unplanned outages or unexpected surges in demand, in order to estimate the expected and minimum capacity margin ahead of time.
New Zealand, however, is dominated by hydro electricity which has highly volatile inflows and limited storage. The first difficulty this poses is that the best inflow data available is the historical record which extends back only to 1931 for the main hydro systems. One cannot be certain that the worst inflows possible are represented in the historical record. Similarly, due to a statistical aberration there could have been several years since 1930 when inflows were lower than 1-in-60.
The second difficulty is that one has to be clear about what 1-in-60 actually means. Does it mean, for example, that one considers each winter in isolation, ignoring the consequences of finishing the year with low storage going into the next winter? Or does it mean that inflows should be projected two years ahead to ensure that a dry spring-summer does not lead to low storage going into winter with a consequent shortage? In these cases, there is an obvious difference between projecting inflows ahead assuming inflow years always follow each other as they have in the past, on the one hand, or on the other hand assuming that any historical year can follow any other regardless of the order they actually occurred in.
The third difficulty is that inflows are uncertain and finite. Even though the capacity margin might be greater than zero in May, for example, there is still the possibility that a shortage situation could develop because a low inflow sequence from that point on through winter would be highly likely to create a shortage at some later time.
The fourth difficulty is that it is impossible to take account of all events that could impact on security of supply when optimising the operation of a hydro reservoir.
The fifth difficulty is presented by the characteristics of our transmission Grid which is often described as being "long and stringy." Most of the hydro generation is in the SI and most of the load is in the NI, but the NI also has most of the thermal plant which must run during dry years to conserve storage in both islands. In a dry year the HVDC link carries flows south, but can only carry so much. As shown in section 6.1 it is a fact that a dry year shortage will affect the SI considerably more than the NI.
The criterion for shortage in a dry year therefore has two parts to it, an operational criteria used for the purposes of managing hydro reservoirs, as simulated in the modelling undertaken for this security update, and a criterion to account for unforeseen contingencies.
6.2.1 Operational Criterion
The need for and use of this criterion explains, for example, why it has been necessary to call for voluntary savings in the recent past even when SI reservoirs were far from empty. It also implies that the dispatch of reserve plant should take place only when there is a high probability of shortage in the future. As this probability increases, one should ideally see increasingly expensive plant running until the reserve plant is dispatched after all other available thermal plant is running hard.
Note that the probability of shortage varies with both storage and time of year. For example, SI storage of 1,000GWh might represent a high risk of shortage in May, going into winter, but a low risk in November when spring and summer inflows can be expected to fill reservoirs over the summer.
Under the new regime governed by the EC, we have assumed that the EC will contract or otherwise work with the industry to ensure that thermal plant runs soon enough when the probability of shortage starts to rise. The dispatch of reserve generation should occur only when the probability of shortage rises still higher.
6.2.2 Contingency Criterion
Because the water value calculations can not take into account all possible contingencies, it would be ill advised to use "empty storage" as the criteria for a shortage. It is an underlying assumption in all of our modelling that it will be planned that some water should always be held back, especially in the large storage lakes in the Waitaki system, in case of significant unforeseen events such as:
- outages of hydro plant in the SI;
- outages of thermal plant in the NI;
- a surge in demand;
- local effects such as constraints within the Grid that make it difficult to deliver power to specific areas of the country, particularly in the SI;
- a dry year worse than any year on record;
- Grid outage.
Figure 7: Unexpected Events Check
In our first December 2002 report, we used the value of 300GWh for SI storage as the indicator of a shortage.31 In this round, two test scenarios were run for the 2006/2007 year with 2.5% demand growth, one with an unplanned outage of 360MW of NI thermal plant in the critical period of a 193232 inflow sequence and one with an unexpected surge in demand of an extra 1% over the same 1932 inflow year. These showed that a buffer of 300GWh is just adequate as shown in Figure 7. However, this takes account of only one unforeseen event whereas we believe the EC may be more conservative and may wish to consider scenarios in which two or more unforeseen events occur e.g. an unforeseen outage and a surge in demand. Hence in our results we report events when storage falls below 300GWh and below 500GWh.
Readers should also note, for example, that in some scenarios SI storage drops below 500GWh late in spring even though Whirinaki reserve generation does not run. This is because demand is lower in spring than in winter and hence there is other plant available to run before Whirinaki. The risk of shortage is also lower late spring than in winter.