• Home
• Energy & Resources
• Electricity
• Reports
• Wind Energy Integration
• Final Report
• 8. A Low Demand Case Study

• Electricity

• Reports

• Wind Energy Integration

• Final Report
• Final Report - Summary
• Media Statements
• Questions and Answers

• 8.1 Discussion
  • 8.1.1 WTG Assumptions
  • 8.1.2 Geographical Dispersion of Wind Farms
  • 8.1.3 Wind Forecasting Assumption
  • 8.1.4 Base-load Assumptions
  • 8.1.5 Conditions for Very High Levels of Integration

Much can be learned by investigating how much wind energy could be supported on the grid during periods when demand is at its lowest, and therefore installed wind generation is at its highest proportion of total generation. Periods with higher demand might support more wind generation but would also increase the amount of time, across any given year, when there is too much base-load generation, leading to excessive spilling of wind or water. Using minimum demand therefore allows us to minimise the impact of this study's lack of assumptions about the relative economics of wind energy.

In 2004, New Zealand generation⁶¹ hit its lowest value for the year of 2,667 MW in the half hour ending at 4 am on 27thDecember. For this study, the North and South Islands were analysed separately because of the nil and limited sharing of regulation and IR, respectively, currently available across the HVDC link. National minimum generation between the islands is also split pro rata on minimum island demand to ensure that transfers on the HVDC link are not counted in either island. This gives adjusted North Island minimum generation of 1,509 MW and 1,156 for the South Island. Adding 2% for growth gives us approximately 1,550 MW and 1,180 MW, respectively, for the current year.

How much wind generation could be supported in the North Island with demand low enough that only 1,550 MW of generation is required to meet demand? Let us assume that it is a warm but windy summer night with wind speeds high enough to give us up to 100% loading on all wind farms.

We look in turn at the key factors which will determine how much wind we can support:

• provision of regulation;
• assessment of reserve risk and provision of IR;
• the need for a marginal station to be the load follower;
• how the system will respond during a fault;
• how grid stability can be maintained including the support of voltage;
• how the SO can manage security of supply on the grid out to the end of pre-dispatch time.

Consistent with the objectives of the study, the analysis is based on a number of assumptions that are, when taken in aggregate, favourable to high levels of wind energy integration. Firstly, it is assumed that future wind farms will consist of either synchronous turbines such as those produced by Windflow Technology or the latest DFIG pitch controlled turbines such as the Vestas V90 which Trustpower intends to add as Stage III of its Tararua wind farm.⁶² These WTGs both provide fault ride-through and inertia to assist frequency management.

Secondly, we assume that the contribution of wind energy is not limited by being dispatched down or off by the SO. Hence they cannot provide IR nor be the marginal station.

Thirdly, we have assumed that wind would displace all other base-load plant⁶³ (except some small embedded stations) so that river flows that would normally be used to generate to meet resource requirements for minimum river flows, for example, would be provided by spilled water.

Finally, we assume that wind farms are geographically dispersed around the country and around each island.

The analysis is represented graphically in Figure 13 and shows that 1,300 MW of wind generation could be integrated during the times of lowest demand.

Figure 13: North Island Low Demand Analysis

→ Figure 13: North Island Low Demand Analysis [21 KB GIF]

Given 2004's peak national generation of just over 6,515 MW, adding growth of 2% gives a peak national generation figure of about 6,630 MW for 2005. If the installed wind energy base were 1,300 MW then this would equate to penetration of 19.6% based on the forecast 2005 year and relative to national peak generation. Assuming a load factor of 40%⁶⁴ this would equate to market share of just under 12% relative to the total forecast generation for the country in 2005.⁶⁵

The analysis for the South Island is shown in Figure 14 and gives a penetration of 930 MW or 14% relative to peak generation and market share of just over 8%, both relative to national figures. In total, over both islands, the low demand analysis gives total installed wind generation of 2,230 MW with penetration of just under 34% and market share of 20%.⁶⁶ Given our assumptions and associated uncertainty in this figure, we will refer to the initial penetration result using the methodology as 35% by rounding to the nearest 5%.

Figure 14: South Island Low Demand Analysis

→ Figure 14: South Island Low Demand Analysis [21 KB GIF]

8.1 Discussion

The analysis is based on a number of assumptions which we now consider in more detail.

8.1.1 WTG Assumptions

If synchronous - synchronised and the latest DFIG pitch controlled WTGs, or WTGs with equivalent capabilities are not predominant with high levels of penetration then they are unlikely to provide voltage support and fault ride-through and may not contribute inertia when frequency falls. Under this scenario, theoretical wind penetration could be much lower in order that the grid could be operated safely and securely.

8.1.2 Geographical Dispersion of Wind Farms

The calculation of the maximum half hourly swing of 26% in aggregate wind farm output, to be covered by the marginal generators, uses overseas data relating to two wind farms that are separated by 200 km. Use of this figure is partly based on our key assumption of geographically dispersed wind farms, and partly an assumption that New Zealand's regional wind speed distributions correlate with each other in the same way as they do overseas.

At present, wind farms are highly concentrated in the Manawatu where 97% of the installed wind capacity is situated. Based on the information in Table 1 and Table 2 this may reduce to 61% by the end of 2008. But, so far, the trend is actually one of significant geographical concentration rather than our assumed geographical dispersion.

Assuming a "worst case" scenario for geographical dispersion, close to 100% of all wind generation would be concentrated in the Manawatu (whether or not it is physically possible for this to happen given the available sites suitable for wind farms.) Initial application of the methodology, with all other assumptions remaining as above, would give penetration of only 15%, less than half of the penetration calculated above. In this case a second regulating station is allowed in the North Island to cope with the large swings in aggregate output observed to date in the Manawatu.

In this case, however, the separate treatment of the two islands now introduces larger errors than before. In reality, with no wind farms at all in the South Island, substantial amounts of load following generation would often be provided over the HVDC link, also allowing Manawatu wind farms to supply up to about 600 MW into the South Island. These two factors could potentially lift wind penetration back up to 23% in this scenario.

It was also noted in 7.9 that large amounts of wind energy in the Manawatu could increase the occurrence of line constraints on the BPE-TKU line and the HVDC link, potentially also limiting penetration to well below 35% unless the constraints could be relieved or managed by further investment in the grid or by restrictions on wind farm output.

While this particular scenario is unlikely, it nevertheless illustrates the potential impacts and uncertainties of a lack of geographical dispersion on the limits to wind integration.

8.1.3 Wind Forecasting Assumption

The initial analysis assumes it is a warm night with enough wind for all wind farms to be at full output. This assumption is probably quite reasonable in respect of a number of summer nights in our hypothetical 2005, but it also implies that wind speed can be predicted accurately enough up to the end of the next day, to give a wind planning margin of zero. This assumption is valid at times of low demand, as there is likely to be a large amount of spare (hydro) generation capacity that can be started quickly if required.

8.1.4 Base-load Assumptions

The assumption that wind could displace almost all other base-load plant on the night of low demand would also have to apply to other periods when demand is only just a little higher. So this scenario could either see increases in hydro spill or spill of wind, affecting the economic viability of one or both of these types of generation.

However, as shown in Figure 15⁶⁷ demand is comparably low for only a small portion of the year. For example, only 0.4% of all half hours in the year have demand in the lowest MW decile.

Figure 15: North Island Load Duration Curve

This, in combination with an average load factor for wind of 40%, means that periods where excess wind creates hydro or wind spill will be relatively few. In addition, as demand grows the capacity of must run hydro is likely to decline as a proportion of total demand.

The impact of base-load competition could in theory create an economic limit on wind integration in the short term, while the proportion of base-load hydro is high, but this limit is expected to be well above current levels of wind penetration. Because the total quantity of hydro electric generation is static, it is also likely to become less significant in the long term as the relative amount of base-load hydro reduces.

Wind, in competition with other forms of generation, will be developed at a pace which is constrained by load growth and the relative economics of wind energy. At this rate it is likely to take 10 or more years to achieve 35% wind penetration (or 20% market share). During that time it is also conceivable that the development of this technology and more favourable economics will allow the limit to be pushed beyond 35%.

8.1.5 Conditions for Very High Levels of Integration

During this study we were not constrained by a requirement to consider how the relative economics of wind generation, including its impact on electricity prices, might influence the limits to wind integration. Our brief was to study what could happen, based on connection, technical and operational factors.

Nevertheless, to put the limits above in context, it is helpful to note that higher limits of wind integration could be consistent with substantial spilling of wind or water in future scenarios in which electricity prices rise significantly above current levels, or the per megawatt cost of WTGs falls significantly. The likelihood of either economic scenario was not considered during the study.

As described in section 10, developments in technology have brought wind farms to the point where they can provide most of the features and capabilities of conventional generation plant. As the more advanced control features described in section 10 become mainstream, the impact of the variability of wind farm output will reduce. These advanced features will allow control over the output of a wind farm within the limits of how much the wind blows. For example, if the wind is blowing at a speed which allows a wind farm to generate at up to 20 MW, the advanced controls could set the output of the farm to any value from its minimum practical output up to 20 MW, under direction of the SO. The greatest limit to the potential of output control will not be how volatile the wind speed is, but how much the wind is blowing on average over an extended period.

Despite the limitations of this initial analysis, it nevertheless indicates that wind energy could achieve levels of penetration and market share that are at least comparable with the higher levels currently achieved around the world, and which would have appeared far-fetched until only recently. This discussion suggests the following conditions would, in combination, support very high levels of wind energy integration:

• technology advances that continue to bring features common to WTGs closer to those provided by conventional generation (including features that contribute to frequency management);
• geographical dispersion of the wind energy sector;
• improvements in the accuracy of wind speed forecasts;
• continued improvements in the relative economics of wind energy;

⁶¹This figure does not include all embedded generation running at the time.

⁶²The ratio used in the analysis is 50:50 although this impacts only on the assessment of large scale wind energy on the requirements for IR. The point is not that these particular WTGs are required, but that WTGs with similar technology or characteristics would be assumed.

⁶³The amount that any generator runs is an outcome of the SO's dispatch process so no-one can guarantee dispatch. For the sake of the analysis we have merely assumed that it is possible that wind could displace all other base-load plant.

⁶⁴Wind farms in the Manawatu are achieving 45% load factors but wind farms in other regions may not, therefore a slightly lower figure than 45% is used in the analysis.

⁶⁵Peak generation is split roughly 57:43 between the North and South Islands, so 1,300 MW in the North Island would represent penetration of just over 34% relative to North Island peak generation.

⁶⁶930 MW in the South Island would represent penetration of just over 33% relative to South Island peak generation

⁶⁷The load duration curve in figure 15 shows the distribution of load from lowest to highest in the North Island for 2004.