Onslow Pumped Storage: an Overview


There has been recent interest in pumped storage in New Zealand in the light of the requirements for transitioning to a low-emissions economy. This post incorporates some results of theses on Onslow pumped storage carried out at the University of Waikato from 2005.

There is a need for recognition that transitioning to a post-gas sustainable low-carbon economy in New Zealand will require a significant increase in energy storage capacity – far beyond notions like Tesla batteries or hydrogen storage. The only practical option is pumped storage, most probably at Onslow in Central Otago. Preliminary evaluations as to whether to proceed at Onslow should start as soon as possible because of the long lead time between project consenting and completion for large civil engineering projects.

It is interesting to speculate as to how things might have been different if the idea of pumped storage at Onslow had been around in 1992. Back then, there was an electricity crisis arising from low hydro lake storage due to low inflows, with almost a sense of desperation about how this might be avoided in future. It is entirely possible Onslow pumped storage could have been operational by 2000, then sold to Contact Energy as part of the Clutha Power Scheme. That would have been a useful asset for Contact Energy because it would enhance their limited storage at Lake Hawea and provide the basis for significant wind power development.

What pumped storage does

Like a battery, pumped storage is a means of energy storage. Power drives pumps to move water from a lower reservoir to a higher reservoir, creating a fixed amount of gravitational potential energy. Power is extracted later by running the water back down, with the pumps running in reverse as generators.

The amount of energy stored depends on the volume of water raised and the height difference between the upper and lower reservoirs. If the height difference is doubled then half the original volume of water is sufficient to store the same amount of energy. Similarly, doubling the height difference means that the rate of water release can be halved while still producing the same power output.

The upper reservoir must be an existing lake or constructed reservoir but the lower reservoir is not necessarily a “reservoir”. For example, a river might be utilised. Some current investigations are taking place in Australia over the possibility of seawater pumped storage, with the ocean as the lower reservoir.

All energy storage systems have some degree of inefficiency. Pumped storage efficiency has improved with turbine development and a modern system could have a to-and-fro efficiency of around 80%. Some further loss may arise from scheme-specific factors such as how far apart the two reservoirs are.

A pumped storage scheme has two components: energy storage capacity and installed capacity. The energy storage capacity is the maximum amount of energy that can be stored, often specified in GWh. The installed capacity is the maximum power associated with generation / pumping, typically given in MW.

For small amounts of energy there are a range of different storage methods available, including compressed air, flywheels, batteries, electric thermal, molten silica, and hydrogen production and storage. All this is specifically with respect to energy storage and release systems. There is of course a considerable amount of potential energy held in a lump of coal or uranium. Different storage systems have different efficiencies. Energy storage for power generation from “green hydrogen” (storing hydrogen produced from water electrolysis using renewable power) is particularly unattractive because of low efficiency coupled with the expense of gas storage.

For storing large amounts of energy there is no option other than pumped storage and no obvious alternative on the technology horizon. The current total global energy storage capacity is made up almost entirely of pumped storage.

Most pumped storage plants around the world have high installed generating capacity and low energy storage capacity. This is because their purpose is usually to provide brief peaks of power to match short-term fluctuations in demand. For example, the Fengning scheme in China will have installed capacity of 3,600 MW, making it the world’s largest pumped storage scheme by installed capacity. However, it could only maintain this power output for a few hours. At the other extreme, some pumped storage plants have sufficient energy storage capacity to operate seasonally.

Pumped storage for New Zealand?

An economic and environmental case can be made for a New Zealand pumped storage scheme with large energy storage capacity. Unlike Norway and Canada, our hydro generation capacity is not matched by a large hydro storage capacity. As is well known, this can lead to upward volatility of our wholesale electricity prices when hydro levels decline.

For example, 2019 did not rate as being an “abnormal” hydrological year but it is interesting to see the electricity price effect of the Southern Alps heavy rains of March 26. This was the same weather event which caused temporary closure of the main West Coast tourist route through washout of the Waiho River bridge.

As seen in the graph, the New Zealand total hydro storage declined through March prior to the flood inflows into the South Island hydro lakes. Wholesale electricity prices over this time were around $200 per MWh. Those buying power from the spot market would have been impacted. This would also apply for major electricity users taking out hedge contracts because the hedge cost would have to take into account the risk of the water in storage becoming even further reduced.

Effect of March 26 2019 rain event on wholesale electricity prices. Prices are median daily values at Haywards node (excluding weekends). Daily storage values provided by NZX.

The sudden increase of southern lakes hydro storage from March 26 caused prices to quickly drop to around $120 per MWh. In other words, if the storage had been at 3,200 GWh from the start of March then the prices would never have been far from $120 per MWh. Equivalently, if the flood inflows had not happened then the prices would have remained at around $200 per MWh.

There was also an environmental impact from the higher prices because they enabled coal-fired power generation.  In the two weeks prior to the flood events the rate of coal burning at the Huntly Power Station was never less than 7,000 MWh per day. On the other hand, however, if a punitive carbon tax had been in place then the coal might not have been burnt at all and the electricity price would have been even higher.

The desired trend in New Zealand is toward an even greater renewable contribution to power generation, with 100% being an aspirational goal. This applies particularly in the longer term as the natural gas reserves are depleted and further exploration for new gas fields is no longer permitted. However, the present situation of weather-related electricity price variation seems no route toward a goal of electrification playing a significant role in a low-carbon economy.

A major increment of hydro storage capacity would seem a logical step toward achieving lower and more stable electricity prices, as well as buffering against dry years. The required increase in storage capacity could in principle be achieved simply by permitting massive lowering of lakes Tekapo and Pukaki if required. This would hopefully be regarded as unthinkable, given the degree of shoreline impact already evident on the southern hydro lakes.

The main South Island hydro lakes are in fact poorly suited to hydro storage from the environmental point of view. The lakes were just the low-hanging fruit for quickly and cheaply creating storage, and environmental aspects did not have such as high profile as now.

The shoreline environmental issues arise from the geological origin of the lakes. The South Island lakes are glacier-created and held in place by soft moraines and glacial tills. Imposing metres of lake level change for seasonal hydro storage in effect results in a mining operation as lake shorelines are wave-eroded and retreat.

A sense of the seasonal water level variation in the southern hydro lakes is gained by comparing the Waitaki River flow at Kurow, before and after hydro development. The pre-hydro Waitaki River had high flows in spring and summer, reflecting melting snow and summer precipitation mostly as rain. Winter flows were low because much of the mountain winter mountain precipitation occurs as snow. However, the main electricity demand is in winter, so the high summer inflows to Lakes Tekapo and Pukaki are now held back to give the needed high lake levels for winter power generation. The end result is reduced seasonal flow variation in the lower Waitaki.

Mean monthly Waitaki River flow at Kurow, before and after upper Waitaki hydro development

Ideally then, new storage would result in reduced power prices while at the same time reducing the seasonal storage variation from the southern hydro lakes. This leads to the possibility of pumped storage at Onslow.

The Onslow setting

Lake Onslow looking south

Lake Onslow is located among gentle hills in a schist rock basin near Roxburgh in Central Otago. It is a small artificial reservoir 8.3 km² in surface area, with a maximum depth of 10 metres near the small dam at the Teviot River outlet. The nature of the topography means there is a considerable increase in lake surface area as dam height is increased. The lake has been raised a number of times since the first dam was constructed in 1888, most recently in 1982 to give the present lake.

The gentle relief and rock nature of the bounding hills reduces risks of rock instability in the event of water storage. It is unlikely there would need to be landside mitigation operations as in the Cromwell Gorge. The schist rock would also provide a more erosion-resistant shoreline compared to the soft glacial deposits around the South Island main hydro lakes.

Small pumped storage schemes could be set up almost anywhere in the New Zealand where there is a hill comprised of sufficiently strong rock to hold an upper reservoir. A major energy user might consider a hill-top scheme if there was sufficiently large variation in electricity price at different hours of the day. Such small distributed pumped storage schemes are of value for providing brief peaking capacity on the scale of a day or so, and could be of particular value in association with wind power development in the North Island.

However, large installed capacities of small distributed pumped storage schemes do not represent energy storage capable of buffering a dry year. That would be like saying that a set of small ponds has the energy storage capacity of Lake Tekapo.

Finding a suitable location for an upper reservoir large enough for dry year buffering is difficult. In this regard, New Zealand is fortunate in having the Onslow rock basin, which is located within pumping distance of Lake Roxburgh. This appears to be the only New Zealand site suited to the required scale of energy storage and has recently gained recognition as a potential dry year buffer toward a national goal of 100% renewable electricity for both normal and dry years.

Energy storage and lake surface area of an expanded Lake Onslow, as a function of the water surface elevation of the new lake

Onslow pumped storage

A pumped storage scheme at Onslow would operate by pumping water up from Lake Roxburgh through a tunnel and into an expanded Lake Onslow. Some of the water would then be run back later to Lake Roxburgh at times when power generation is required. The essential civil engineering would comprise a long 24 km rock tunnel and an earth dam at the Teviot River Onslow outlet. The installed capacity could be around 1,300 MW. Total cost would be in the range of 3-4 billion dollars, depending on scheme size.

An earlier paper considered a particularly large Lake Onslow, with maximum water level at 800 metres above sea level and extending over into the Manorburn reservoir (located to the north on the map). Coupled with a large operating range, that would be the world’s largest pumped storage scheme by energy capacity. However, its size and potential environmental impact raise both environmental and economic concerns. The paper nonetheless did provide a service in identifying a location for the purposes of large-scale storage, should 10,000 GWh be required some time in future to ensure 100% renewable electricity generation.

The Onslow-Manorburn basin. Sourced from LINZ

A recently-completed PhD thesis at the University of Waikato simulated the operation of a somewhat smaller scheme, with an expanded Lake Onslow operating between 720 and 780 metres above sea level.  These specifications were just for the purposes of a thesis study to evaluate the possible influence of a large pumped storage scheme on grid operations. A 60-metre permitted operating range is still extremely large and this would cause the lake surface area to almost double between maximum and minimum storage. An indication of the extent of the fluctuation is seen in the figure below.

Possible extent of a raised Lake Onslow. The outer boundary of the grey region shows the extent of the lake at 720 metres above sea level. The outer contour shows the extent of the lake at 780 metres above sea level. Scale is in kilometres. Dashed line shows the extent of the dam crest for a lake at 780 metres.

A useful aspect of the Onslow site is that a smaller pumped storage scheme could be constructed to still give some worthwhile increase in energy storage capacity while maintaining the full 1,300 MW of installed capacity. For example, the present 14.5 metre operating range of Lake Pukaki could be translated to a raised Lake Onslow, to operate between 720 and 734.5 metres above sea level. This scheme would have a maximum lake extent only slightly greater than the grey area in the figure above. The new lake surface area would vary between a minimum of 40 km² to a maximum of 47 km².  Onslow energy storage capacity in this configuration would be about 1,000 GWh, which would increase national energy storage capacity by close to 25%.

Onslow pumped storage does require a considerable increase beyond the present Lake Onslow surface area to give energy storage capacity and to avoid major water level fluctuations during operation. This is a major environmental change but at the same time gives opportunity for creating a significant environmental asset which could be of as much importance as the infrastructure gain from the pumped storage scheme itself.

This arises because in Otago there has been a loss of about 90% of the region’s wetlands. Today even the loss of a small upland Otago wetland is cause for concern. In fact, the present Lake Onslow also contributed to wetland loss because the lake submerged Dismal Swamp – its very name reflecting the distain held for wetlands in earlier times.

Keeping the present area of Lake Onslow as open water for recreation, an extensive floating wetland in excess of 30 km² could be constructed over most of the expanded lake. That is, a new Otago wetland larger than the area of Lake Dunstan. The requirement would be for an extensive and intricate network of reeds and narrow waterways, the latter planned carefully to enable passage of eco-tourism excursions from Roxburgh. An ecological advantage is that this wetland environment would be robust against climate variation and would not dry out in a drought. The entire Onslow system could be surrounded by kilometres of predator-proof fencing in support of wetland birdlife.

Constructing an environmental asset of this type as an end in itself would normally be impossible because of cost factors. However, in relative terms the wetland cost would be not be so great when compared to the total pumped storage scheme cost.

Environmental consideration would also need to be given to the new earth dam at the Teviot river outlet from the lake. For a lake level extending to 734.5 metres above sea level there would be a dam height of around 55 metres. However, the Teviot Stream outlet is a narrow valley so the embankment height would be much less than 55 metres over most of its extended length.

It is important that the dam should not look something like a smaller version of Benmore. Nature gives some guidance here with respect to Lake Waikaremoana. Driving up to Waikaremoana gives no sense of a dam which is distinct from the surrounding countryside. This is because the “dam” is actually an ancient landslide that was soon covered over by native forest. In a similar way, the Onslow dam should be constructed to be indistinguishable from the Central Otago landscape. That is, it should be overlain with constructed ridges, schist rock outcrops, and tussock cover.  There would also need to be minimal disruption to present local hydrology. In particular, water release to the Teviot River from Lake Onslow should remain as it is now.

Onslow operation

If it were to be constructed, the initial lake-raising period at Onslow would be a time of net energy sink. For example, the amount of energy required to raise Onslow from its present level to a minimum operating level of 720 metres above sea level would be about 4000 GWh, which is roughly the same as the New Zealand national storage capacity. Put another way, it would take 167 days to pump up the required volume of water by pumping continuously at 1000 MW.

In reality, the required time would be considerably longer because sufficiently cheap power for pumping will not be often available. One possibility for finding the required energy could be national electricity conservation campaigns from time to time, with an incentive being future electricity prices lower than they otherwise would have been.

Another factor requiring a slow filling process is that water loading of new large reservoirs is done by increments, with extended waiting time after each water volume increment to check for any enhanced seismic activity. There are no mapped faults under Lake Onslow but such precautions are standard.

Even though the filling time may be over an extended period, the scheme could meantime have active use for buffering short-term wind power fluctuations.

Once the minimum operating level is reached then the Onslow scheme could operate to moderate the seasonal variation of the main southern hydro lakes. For example, high spring and summer inflows to Lake Pukaki could now be released from the lake and not held back for winter generation. The excess spring and summer power from the Waitaki power stations would be used to pump water up to Lake Onslow from Lake Roxburgh. Winter power generation would then be in part from the Waitaki power stations and in part from water release from Onslow. This is only a rearrangement of storage and there is no transmission line upgrade required north of the southern hydro lakes.

To summarise this seasonal process – water is released from a lake at the top of a hill to generate power to pump water up into another lake at the top of another hill. The practicality of this apparent energy irresponsibility arises from the limited storage capacity of the present lakes. From time to time, high water levels in the hydro lakes lead to spill losses when flood inflows pour into a lake which has no further storage space. Such spill losses do not happen often but there can be considerable lost generation opportunity when they do. This applies particularly for the Waitaki hydro scheme where there can be simultaneous spill at all power stations.

If water is instead released from the Waitaki lakes so that their water levels remain more often near mid-range, there will be lake storage on hand when major inflow events occur. That inflow water is now available for later generation rather than lost to spill. At the same time, there is the shoreline environmental advantage in reducing seasonal water level fluctuation in the hydro lakes. Also, the Waitaki River flows would revert back more toward their pre-hydro state so more summer water is available. For example, there might be increased river diversions for irrigation while there is still increase in the Waitaki summer flows. Finally, there would be some useful flood peak reduction in the lower Waitaki.

Much of the PhD study was concerned with simulating this mode of operation coupled with the large 60-metre hypothetical water level variation at Onslow. One thing that emerged from the simulations is that in any one year there was much less than 60 metres of water level variations. As far as we could estimate, the additional power yield from reduced spill resulted in a net power gain after allowing for inefficiencies in the Onslow pumping operation. That is, the net effect of Onslow pumped storage in this mode of operation is to create a small power station as a long-term average, mostly arising from more power generated from the Waitaki power stations. Contrary to popular belief it is in fact possible to generate power by pushing water uphill.

The prime purpose of Onslow pumped storage would be to provide energy reserve to buffer through dry periods like 2007-2008. It would of course be uneconomic to construct a 24 km tunnel just to extract some more power from the Waitaki stations or smooth short-term flucations in wind power. However, once constructed, it is highly probable that the Onslow turbines would be operational every day in support of wind power.

Reality check

The thesis simulations were carried out as an academic exercise that takes no note of the present structure and ownership of New Zealand generation assets. For example, Lake Hawea is controlled by Contact Energy, Lake Pukaki by Meridian, and Lake Tekapo by Genesis. It would seem unlikely that these competing entities would cooperate and agree to restrict their water level operation after spending billions of dollars on an engineering scheme that would have the end effect of lowering the price of the electricity that they sell. However, company-centric attitudes may become increasingly antiquated in an environment where climate change is seen as an increasing threat.

One option might be for the government to give advance notice that N years into the future there will be a heavy tax imposed on high levels of the existing hydro lakes. If a carbon tax is imposed against coal and gas-fired generation on environmental grounds, then it might be justifiable to also impose a lake level tax on environmental grounds. The prospect of losing control of lake storage could then give incentive to seek alternative storage options. However, imposing such requirements seems unlikely. An even more drastic (and even more unlikely) option would be to recreate the old NZED and re-establish integrated control of all New Zealand hydro lakes. Whatever the mechanism, it would be helpful to the nation if the next hydro power scheme was simply an increase in mean power generation from the Waitaki and Clutha schemes, arising from Onslow seasonal pumped storage operation.

Transition to low-carbon

Given the above comments, it might seem that constructing large-scale pumped storage has potential in theory but might be difficult in practice. However, there has been a significant development with New Zealand’s signing of the Paris Agreement and making a commitment to transition to a low-carbon economy. A major component of the transition is a shift to electrification in transport and industry coupled with increased use of renewables, particularly wind power. The requirement here is for increased buffering against calm/dry times by using enhanced storage.  Some early indications of what may happen in future without extra storage was seen in the high wholesale electricity prices in the second half of 2018 when low lake levels coincided with temporary reduction in gas availability. Any “just transition” to a future low-carbon sustainable economy could hardly be regarded as “just” if is marked by high electricity prices and power shortages.

Interestingly, the requirement of increased energy storage as part of the low-carbon transition against climate change had hardly gained public mention in New Zealand, other than a call for further investigation by New Zealand engineers. The New Zealand government’s announced $20m energy research fund seemed to avoid seeking to overview applications of established technologies in favour of hoping for some silver bullet from “organic photovoltaics, super conductors, nanotechnologies and inductive power”. This can be contrasted with the South Australia Government’s announced $50m fund to investigate a whole range of energy storage possibilities, explicitly including pumped storage. It is pleasing, therefore, that pumped storage has now been recently recognised as the most cost-effective means of achieving 100% renewable electricity generation in New Zealand, through provision of the Onslow buffer against dry years. However, a significant cost factor in this context is the need for transmission line upgrades, which would have to be weighed against the desirability of removal of all fossil fuel power generation.

The role of wind

Given that there is sufficient incentive created to expand wind energy to the required degree for the low-carbon transition, there will be some important issues arising.

Importantly, wind and rain have a degree of correlation. Strong westerlies can bring both wind and rain. A sequence of anticyclones brings neither. Suppose at some time in the future the gas supplies have been depleted, coal is no longer permitted, and the transition to a wind and hydro-based electrified economy has been completed – but without additional energy storage capacity.

Superimpose a dry calm winter and the hydro storage from the previous summer will deplete rapidly because there is now greater electricity demand than before, but still the same hydro storage capacity. Power restrictions such as electric vehicle carless days would be unavoidable. This scenario will not necessarily be offset by building an over-capacity of wind power. Even if the present hydro lakes are full at the start of an extended calm/dry period, having a large number of wind turbines producing minimal power will not reduce the rapid rate of water level fall toward emergency levels.

The other extreme is not helpful either, given the present energy storage capacity. Suppose there is an extended period of wet and windy weather. It might be said that this is no problem for wind power because all that is required is to reduce the rate of lake water release for hydro generation and simply utilise wind power as it is created from wind turbines. However, this means that the hydro lakes would rise somewhat more rapidly because increased lake inflows from the rains are coupled with reduced lake outflows resulting from wind power generation. This could create later hydro spill and lost generation opportunity. From a commercial point of view, a generating company would be unlikely to expend capital on a windfarm if a portion of their income from wind power is offset later in the year by spill over their hydro stations. This issue was noted as far back as 2009 by an Otago Daily Times editorial but is particularly relevant now with the intention to significantly increase the wind power component in New Zealand as part of the transition to electrification.

Finally, there is the short-term intermittency of wind power. This arises from the spatial association of wind speeds so that, for example, a sudden drop of wind speed will impact a considerable number of different wind turbines at about the same time. This requires rapid ups and downs of power output from other sources to offset the wind fluctuations. There is a limit to how much such buffering can be supported by our existing hydro stations and some in the industry suggest we are almost at “peak wind” now.

All these issues can be mitigated by significant new pumped storage energy storage capacity coupled with a large generating capacity– which almost certainly means Onslow pumped storage in the absence of any other realistic options. Dry winters could be offset by maintaining large reserve Onslow storage. The issue of wind-driven hydro spill would similarly be mitigated by having the large storage capacity at Onslow, coupled with the fact that Lake Onslow is not supplied by major river inflows subject to flooding. The intermittency issue would also be offset, at least for nearby wind farms, by the large installed capacity at Onslow. This is because every MW of new installed wind power requires about 0.5 MW of new hydro buffer. Having a 1,300 MW installed capacity at Onslow would thus enable 2,600 new MW of South Island wind power. This would be a valuable contribution to the low-carbon economy, although (as previously noted) there would need to be a grid upgrade to allow for the increased northward transfer of wind power.

All comments in this post are based on unfunded University of Waikato studies and do not signal intention by the New Zealand government or other agencies to construct  pumped storage at Onslow. The July 16 response from the Minister of Energy to the ICCC Electrificiation Report is that agencies able to carry out further investigations into New Zealand pumped storage will be identified and announced before the end of 2019.

I would like to recognise the work of my research students Sarah Bear and Mohammed Majeed. I would also like to thank Bryan Leyland for his contributions over the years, as well as the supportive comments from the audience at the New Zealand Electricity Engineers 2006 conference*. Any errors remain my own.

*Bardsley, W.E., Leyland, B., and Bear, S. (2006). A large pumped storage scheme for seasonal reliability of national power supply? Electricity Engineers Association Conference, Auckland, June 16-17.

Earl Bardsley, University of Waikato, Faculty of Science and Engineering