The notion that storage can somehow make the variable renewable sources — principally wind and solar — more attractive, is widespread. The implicit assumption is that storage will enable wind projects to deliver steady power, which, in turn, will make them more valuable.
There is a grain of truth in this assumption, but the difference in value between "steady" and "variable" power is generally fairly small. In some instances it may be sufficient to make a storage scheme worthwhile, but that is probably the exception rather than the rule. "Storage," proclaimed an august body a little while ago, "can transform the economics of the renewable energy sources." Presumably the writer meant "the variable renewables", but, apart from that, the thinking is woolly. The economics of anything can be transformed if you throw money at it, but that rather begs the question as to whether you will end up out of pocket or better off.
However, storage has a role to play in electricity networks with variable renewables. California’s Independent System Operator recently published a discussion paper that describes how its system might look by 2030: "By 2030, California gets more than two thirds of its electricity from non-fossil, non-nuclear resources. Electric supply revolves around the variable output of wind and solar resources, with gas generators running mainly to fill gaps in clean power supply."
It goes on to suggest that renewables will "supply an increasing share of reliability services, including frequency response ... regulation and spinning reserves". Storage is mentioned in various places, including a role for "oversupply of midday solar generation ... by a combination of measures, including exports, market bidding by renewable generators, storage and use of electricity in transportation and buildings."
Early adopter… Duke Energy added 36MW battery storage to its 153MW Notrees wind farm in Texas in 2013
An even more ambitious scenario has been examined for Australia. A team from the Australian National University considered a case where wind and solar would contribute 90% of the electricity supply by 2030 and concluded this was feasible with the aid of various measures, mainly the building of pumped hydro storage. Various scenarios were examined, with the requirement for pumped hydro varying from 14GW to 28GW. While this would present no problem in Australia, it might not be practical elsewhere. The report estimated the cost of balancing — often a contentious issue — at A$24-48/MWh (US$19-38/MWh), depending on the scenario.
The levelling myth
There is a role for storage in electricity systems of the future, but not necessarily for "levelling". One difficulty with transforming variable wind power into "steady" power is that it can only be achieved to a limited extent.
On a local or national scale, any lulls in wind-power generation can be too long to enable the power deficits to be plugged by storage — the required storage devices would simply be too big. More than 40 years ago, a study by the electricity authority in the UK concluded that "150-hour storage is quite inadequate to smooth the output of wind generators".
The difficulty can be illustrated by looking at the power output from wind in Denmark. The graph below shows the output from wind farms in Denmark for a ten-day period in March 2016.
At the end of that year, Denmark had around 5.1GW of online wind capacity, and the average output over the year was 1,454MW. During the ten-day period shown in the graph, the output was below the average for 230 hours — all but about 10 hours at the end of the period. Providing sufficient storage to enable the power to be made up to the average level for that amount of time is out of the question. The energy deficit — relative to the average — over the ten-day period was 246GWh. That is nearly 30 times the capacity of the largest pumped storage hydro facility in the UK and would cost about $37 billion, assuming suitable sites could be found.
Storage may only have a limited role in smoothing the output from wind farms, but it can help manage electricity systems as they become increasingly complex and encompass more variable renewables. Batteries, in particular, are developing rapidly and are likely to have a bigger role.
Wind energy is not a threat
It is important to note that electricity networks are used to managing large swings in demand and the addition of wind energy (which can be regarded as a negative load) into the generation mix does not necessarily cause big problems for system operators. There is a tendency to think that the uncertainties in wind generation must be matched on a one-to-one basis with additional spinning reserves, but this is not the case. There are many fluctuations in consumer demand, and these can mask the wind fluctuations.
The addition of 10GW of wind power to a system with a peak demand of 50GW (roughly the size of the UK network and Duke Energy in the US) increases the average uncertainty in the supply/demand balance, one hour ahead, by about 100MW. Most system operators would cater for this by increasing the extra short-term reserve by about three times this amount, or 300MW. More reserve is needed on a four-hour time horizon, and, according to estimates by UK grid operator National Grid, the reserve requirement will rise from 4GW to 7.5GW as wind capacity grows towards 30GW. However, around 1GW of this extra requirement stems from the needs arising from the construction of larger nuclear-power stations.
Storage might have a role in providing this reserve, but it would be in competition with conventional fossil-fired generating plants, open-cycle gas turbines, demand-side management, or renewables themselves.
The use of batteries is becoming more common as prices have come down (pic: Duke Energy)
National Grid has also highlighted the many possible uses of storage in the 21st-century electricity network. These include frequency management, voltage control, system restoration (in the event of blackouts), constraint management — avoiding the need for wind plant output to be curtailed, and deferring network reinforcements. Batteries are well suited to provide many of these services, since power can be provided almost instantaneously. National Grid recently issued a tender for "enhanced frequency response", and the successful bidders all proposed using batteries. The specification stated that a response is required within one second, which is well suited for batteries, but tough for conventional plant.
As storage technologies have high capital costs — mostly in the $1,000-1,600/kW range — they need to be used intensively to be worth the investment. Storage is a generation technology and has electricity-cost components similar to those of coal and gas. It incurs operation and maintenance costs, as well as a fuel cost for the electricity used for charging. However, the load factors of storage systems are inevitably modest because systems must spend at least half of the time charging.
The graph below shows the extra costs added to the cost of the electricity used to charge a storage device. If the device costs $2,000/kW, can be charged with electricity at zero cost, and can achieve a 30% load factor, it will need to recoup $100/MWh from electricity sales, assuming a weighted average cost of capital of 8% (middle line). If the cost of the electricity used for charging is $30/MWh, it will need to recoup $137.5/MWh for the electricity sold — the extra $7.5/MWh takes account of a typical "round-trip" efficiency of 80%.
Historically, many storage systems were built for "arbitrage" purposes — assimilating energy from inflexible nuclear plants at night (when electricity prices are low) and discharging energy during the daytime peaks (when prices are high). These were mostly pumped-hydro storage systems, but much of the current interest in storage centres on the use of batteries, the costs of which are falling rapidly.
Batteries are less suited to arbitrage, due to their limited energy-storage capabilities, but they are being used for a variety of system services, for which there is a growing demand as the nature of electricity systems is changing, with more small-scale generating systems, many using renewable-energy sources. System services mostly charge electricity prices above wholesale prices, and income can often be generated from several services — a process known as stacking. However, storage faces competition in several of these markets. Demand-side management, in particular, is growing, with sophisticated management systems being used to aggregate contributions from small-scale participants.
The fact that the capital cost of most storage devices is roughly similar to that of wind explains why dedicated storage — a storage system linked to a particular wind turbine or wind farm — is generally unlikely to be economic. The storage load factor, at 40% or less, is similar to that of a wind farm, so the addition of dedicated storage would double the electricity-generating cost of the wind farm, if the storage element had the same rated capacity as the wind farm. However, there is considerable interest in linking modest amounts of storage to wind farms in remote areas, on island systems and to consumer PV, where the economics may be more attractive.
Types of storage
There are numerous types of storage and these can be grouped into four major categories:
Mechanical — this definition covers technologies that use a conventional generator to produce electricity.
Pumped Hydro storage is the most widely used, with about 165GW worldwide. Unit sizes are up to 3GW, with charge/discharge times up to around ten hours. Round-trip efficiencies are around 80% and the units can respond within about 30 seconds.
Compressed-air storage follows some way behind, with about 650MW worldwide. The biggest unit size is 290MW, but larger systems are planned. The compression process heats the air, which means the containment needs to be insulated. If underground caverns can be used, costs are lower than if containment needs to be built. In the generation phase, the air needs to be heated and typically is used in a gas turbine. Round-trip efficiencies are 50-75% and response times similar to those of pumped storage.
Flywheels tend to be smaller systems, with unit sizes up to 20MW, adding up to about 850MW worldwide,. Discharge times are generally one hour or less, although the response time is rapid.
Batteries — there are more than a dozen types of batteries and total global capacity is growing rapidly. Both the Californian and Australian studies, cited earlier, note that the costs of batteries are falling and that they are likely to contribute to the high-renewables electricity systems of the future. Their principal role seems likely to be providing network services.
One of the largest units is a 100MW lithium-ion type in South Australia, which will be used to stabilise the local grid after a fault until backup generation comes online. It can sustain 70MW for ten minutes or 30MW for two hours, and will be used for load shifting by the owner of the 315MW Hornsdale wind farm.
The drawbacks of some battery types are "self discharge" and a limited capability in terms of charge/discharge cycles. Round-trip efficiencies are good, and their main advantage is the ability to respond quickly and provide power when required.
The batteries of electric cars may provide an attractive opportunity to absorb surplus wind power at night, when consumer electricity demands are low, and cars need to be charged. The advantage of using electric cars for storage lies in the fact that the cost of the storage would not be borne by the electricity system, but by the car owners. Using battery-powered cars might, initially at least, be a "one-way" storage application, since there are technical issues associated with feeding electricity back into the grid. Also, car owners might be reluctant to connect up during a shopping trip if there was a risk of the battery being drained on their return.
Thermal stores — these include molten salt (around 2.5GW worldwide), ice or chilled water (250MW). Rated outputs up to 10MW; discharge times up to four hours.
Storing energy as hydrogen is still in the early stages, with Enertrag’s hybrid power plant in Germany one example (pic: Enertag)
Power to gas — there are two options under development, both using hydrogen produced by electrolysis from surplus electricity. The hydrogen can be used directly, either by injection into the gas grid, or by vehicles with fuel cells, or it can be returned to the electricity network via a fuel cell. The latter option would, however, have a low round-trip efficiency.
Alternatively the hydrogen can be reacted with carbon dioxide to form methane, possibly for injection into a gas grid. The "hydrogen economy" has been under discussion for some time, but, apart from a few demonstration plants, has yet to become established. Most studies assume that the prime user of the hydrogen would be road transport, although a fuel-cell powered train has recently been unveiled.
There is a broad consensus that storage will grow rapidly over the next 10-20 years, but not everyone thinks it is on the verge of a breakthrough. According to Energy UK’s Pathways for the GB electricity sector to 2030: "Electricity storage is widely regarded to be the single most important technological breakthrough likely to happen over the period to 2030 and a complete ‘game changer’ in the way that the power system operates. Views [from a consultation] varied on when storage would be commercially viable at either a consumer or grid level, but many respondents argued it would be commercial at a distributed level within three to five years. One company said that it would ‘always remain ten years away’."
The National Grid’s Future Energy Scenarios suggests UK storage capacity will rise from its present level of 2.7GW to between 5-10GW by 2040.
Building blocks — GE’s Reservoir platform for renewable energy
GE has announced a new energy storage platform utilising its internet of things (IoT) technology, with a 1.2MW/4MWh unit as its first product.
The Reservoir platform will typically be used to complement wind, solar and thermal power applications, alongside generation, transmission or distribution assets. GE says it will help power generators increase renewables integration.
The platform will use GE’s IoT Predix and Edge control technologies to "provide data-driven insights that help energy operators enhance their systems".
Eric Gebhardt, GE vice-president and strategic technology officer, says: "Solar is a fantastic resource; wind is a fantastic resource. But they are not necessarily producing the energy when it is needed by the end consumer, so this can be handled by the grids.
"The penetration levels today are getting higher, but it still needs to be managed. Today, electricity has to be produced and consumed at the same time. This allows a buffer in- between for that."
The first product of the new platform to be unveiled is a six-metre 1.2MW/4MWh Reservoir storage unit (pictured).
According to Gebhardt, users can stack up as many units as they want — "like Lego blocks" — to handle more energy.
The company also claims the 1.2MW/4MWh unit will be capable of extending battery life by 15%, increase annual energy output by 5%, and would be easy to install.
GE says it has a pre-launch commitment for the platform of 20MW (80MWh).
The company will target the product at "places that have high renewables penetration", says Gebhardt, specifically citing California and Arizona in the US, Australia, and "parts of Europe".
GE has introduced storage technology before. Most recently it unveiled a 41MW battery system to be built in the UK.
Gebhardt points out that the costs of storage is continuing to fall. "It is now an economic way to go ahead and manage the grids."