It is now widely accepted that the race to decarbonise the world's power systems, is actually a race to provide suitable and sufficient energy storage solutions. Ultimately, generating enough renewable electricity won't be the limiting factor, but rather how available (dispatchable) this clean energy is when we need it.
Energy storage bridges gaps between supply and demand in the electricity grid. These gaps can be as short as a few seconds or extend to weeks in cases of deep-winters, droughts or heat waves. Below in figure 1 are the various types of energy storage plotted on a continuum of duration. Included are both technologies commercially viable now, as well as those in the development pipeline that are showing the most promise.
Short duration storage from seconds to minutes helps to keep the frequency in the grid stable, by absorbing over-frequency and boosting active power when there is underfrequency. Aluminium smelters can, and do, provide frequency response services in many markets, however, grid level batteries with their unrivalled speed of delivery are currently ideally placed to provide these services. There are other emerging technologies such as supercapacitors, super conducting magnets and even flywheels that potentially show greater round-trip efficiency than batteries, which may also boost provision of these grid services in the future (see figure 2).
Energy storage for periods from minutes up to four hours is required to help reduce peak-net-load, when demand outstrips supply. While grid level batteries are increasingly being commissioned to provide this response, their capacity is limited comparative to their capital cost, especially when compared to an EnPot enabled aluminium smelter (see Cost Comparison of Low C02 Peak Net Load Reduction).
It should be noted here that while an aluminium smelter (without EnPot) can modulate down for a limited period (1-2 hours), it does so by incurring penalties of process instability and an "energy deficit", that must be replenished in a subsequent "recharging" period. An EnPot equipped smelter on the other hand, suffers no such penalty (see Energy Modulation Window Over Time), and therefore the time frame of modulation can be extended out to beyond 3 weeks.
When it comes to storing large quantities of electricity for durations over four hours, there is a distinct lack of commercially viable options. Pumped hydro storage is the main method used and the benchmark, however the production of methanol (power to liquid), synthetic natural gas, and hydrogen are being also being proposed. The capital cost of infrastructure of these systems, combined with recharging time and cost, and the round-trip energy losses, significantly add to the overall cost of electricity (see figure 3).
Energy storage typically consumes electricity and stores it in some other form, then re-generates it and feeds it back to the grid when needed, making a profit from the price gap between the buy and sell prices (energy arbitrage). The ratio of MWh of energy put in, to MWh of energy retrieved from storage is the "round-trip efficiency" (also called AC/AC efficiency), and is expressed as a percentage. These round-trip energy losses need to be factored into operating costs and final calculations of energy arbitrage.
Some short duration systems (like supercapacitors) have very high round trip efficiencies of approximately 95%, and most forms of four-hour batteries are in the 75-85% range. When it comes to long-term energy storage however, only pumped hydro has a round-trip efficiency of over 50%.
The DSR of an EnPot enabled aluminium smelter on the other hand, has a comparative round trip efficiency of 100%, as all of the electricity liberated by the smelter remains available in the power system. Adding further to the cost advantage of round-trip efficiency, an EnPot enabled aluminium smelter does not require any 'recharging' when returning to normal operations (see Energy Modulation Window Over Time).
All other forms of long-term energy storage require significant time to recharge, thus limiting the time between repeat cycles.
While the costs of renewable energy generation have plummeted over the past decade and are set to decrease further, conversely the costs of balancing the grid to maintain reliability go up as the penetration of renewables generation increases.
Balancing costs make up a portion of the overall cost of electricity. When the costs are plotted against penetration rates (renewable energy fraction) the resulting curve dramatically steepens as the power system nears 100% renewable energy.
For example, in Australia it has been estimated that the present cost of balancing the National Electricity Market (NEM) is low because the renewable energy fraction is low, however the costs will rise steeply once the renewable energy fraction rises above 75%.
Some experts argue that the ratio "MWh of storage required" to "MWh of generation capacity" reaches at least 1:1 as the power system approaches 100% renewable generation and the cost curve becomes exponential, therefore the gap between supply and demand can no longer be bridged by energy storage alone.
To put it more simply, the electricity grid would need to either switch off non-critical users, or 'buy' power back from the users through DSR mechanisms.
The compelling attribute about DSR is that it comes at low (or no) infrastructure cost to the grid, but provides the most valuable MWh.
Additionally, other forms of energy storage are unproductive when not being used to deliver power to the grid, whereas a modulating aluminium smelter is economically productive before, during, and after, providing its DSR services.