Considering that 90% of energy usage takes the form of combusting carbon-based fuels, it should come as no surprise that most energy storage is fuel storage. This makes perfect sense whether the fuel is destined to be used directly for some particular application, or to generate electricity. Coal and hydrocarbon distribution and storage are mature technologies. We know how to dig the stuff out of the ground, how to build hopper cars that deliver coal to powerplants, and how to build petroleum tankers, refineries, pipelines, & storage tanks. Natural gas pipelines are also well-understood, although large-scale distribution of liquified natural gas (LNG) is still fairly new. These activities involving fossil hydrocarbons account for the vast majority of energy storage today. However, with fossil hydrocarbons projected to decline eventually – and with many climatologists who track global warming hoping they will decline faster in importance than just because of resource depletion – it makes sense to look at storage of energy in other forms.
If the world wants to wean itself off fossil fuels, it needs to look at ways of storing electricity. Here we survey some of the more promising methods for large-scale storage and recovery of electricity. On-grid storage will become increasingly necessary for a variety of reasons. If the grid of the future will utilize nuclear as the predominant form of generation, storage will be necessary because nuclear reactors are not readily dispatchable. If what we now call “alternative” technologies such as wind and solar emerge as the predominant technologies, then matching supply and demand will be even more challenging and will require even more storage.
Storing water to generate electricity can take one of two basic forms. The simple case doesn’t involve pumping at all. Consider a reservoir behind a hydro dam: If the utility allows water levels to rise in off-peak periods and drop during on-peak periods, then it has achieved water storage without pumping. This allows energy to be captured from the water as efficiently as operating the plant at a constant output. This is possible wherever hydro dams are found, but isn’t suitable for waterfall or “run-of-the-river” hydro plants that do not have a controllable reservoir. “Pumped water storage” usually refers to scenarios where, absent a hydro dam, water is pumped uphill to a reservoir during off-peak periods, and allowed to run down again during the peak, driving a hydro turbine at the bottom in the usual manner. The gravitational potential energy E stored in the water is E = mgh, where m is the mass of the water, g the acceleration due to gravity, and h the height through which the water falls to the turbine. The power needed to pump the water uphill need not have been produced hydroelectrically. Pumped storage could be used in conjunction with nuclear plants, for example. To be truly useful, the reservoir needs to have enough capacity to handle an entire daily peak, typically six hours of the peak portion of demand.
Pumped water technology allows recovery of about 75% of the energy consumed in pumping the water uphill, and is currently the most cost effective form of mass power storage. There is over 90 GW of pumped storage in operation around the world, which is about 3% of instantaneous global generation capacity. The main difficulties with pumped water storage are two: First, the 25% energy loss is a fairly high energy cost to pay for getting storage. Second, suitable sites are not easy to find, because creating a large reservoir requires either an elevated natural lake, an elevated dammable ravine, or a plateau on which a reservoir can be dug.
The concept here is to compress air into a large reservoir during off-peak periods, and then allow the air to expand and drive a pneumatic turbine during peaks. Remember that compressing gas costs energy which ends up increasing its temperature, while expanding gas consumes energy for the expansion but cools the gas. This follows from equations of state for gases, such as the ideal gas equation mentioned in an earlier post, PV=nRT. Here the “P” is the pressure, “V” the volume, and “T” the temperature of the gas. “n” is the amount of gas (the “molality” or number of molecules), and R is the ideal gas constant. When compressing a fixed amount of gas, the work that is done increases the energy of the gas but decreases its volume. Since n and R don’t change, if V decreases then P must increase to make up for the reduced volume, and T must increase to make up for the increased energy – unless, of course, the heat is allowed to escape. However, if the heat is allowed to escape after compression, then that amount of energy will be missing during expansion. A compression and storage process that retains all the heat is called adiabatic, while one that allows heat to escape during compression so that the temperature doesn’t rise beyond the ambient is called isothermal. In between these are diabatic processes. The best case scenario for compressed air energy storage is an adiabatic facility. That ideal is impossible to achieve in real life, although it may be possible to come close enough to call a practical plant “adiabatic” without too much shame.
Being more complicated and less efficient than pumped water storage, compressed air storage is not widely used by electric utilities. The earliest large compressed air plant was at Huntorf in Germany, with a capacity of 290 MW. Another facility is in the USA at McIntosh Alabama with a capacity of 110 MW. Both are diabatic plants. It is estimated that an adiabatic plant could achieve 70 % efficiency vs a little over 50% for diabatic plants.
When batteries are used for stationary electric power storage, the constraints are somewhat different than for mobile applications. For mobile systems, energy density and specific energy are of paramount concern, but it’s also very important to achieve low maintenance, and the ability to operate in cold weather. Batteries used for on-grid power storage might benefit from these characteristics too, but more important still are cost and efficiency. For this reason, the battery chemistries being explored for stationary batteries are very different.
The main attraction of the lead-acid battery for stationary storage (and for many other applications) is that it’s an inexpensive technology, and with strong peak output at least for short periods. Wikipedia has a good summary of these batteries’ performance. The specific energy and energy density are unimpressive at 0.11 to 0.14 MJ/kg and 0.22 to 0.27 MJ/litre respectively. The charge-discharge cycle efficiency varies between 50%-92% depending on various factors such as the quality of the battery, its age, how often and how deeply it has been discharged, and how rapidly it’s being charged or discharged. The batteries can last up to 800 cycles if they are never deeply discharged. The self-discharge rate is 3 to 20 %/month, so this technology isn’t very good for long term storage.
Another candidate is the sodium-sulphur (NaS) battery. There is a test project currently underway by Ecel Energy with such a battery system. Sodium metal is corrosive and explosive on contact with water, and the batteries need to be kept at temperatures well above room temperature, which would limit this battery type’s appeal for automotive use. However, it does have a relatively high specific energy of 0.325 MJ/kg and the materials aren’t rare. In the Ecel test setup, an 80 tonne battery has a capacity of 7.2 MHh (26 GJ), and offers a 1 MW throughput, enough to deal with several hundred homes. The efficiency is not disclosed. The cost of the project is high at 1M$, but remember this is a “one-of” prototype so far with staff looking after it. It is not clear how deeply the NaS battery can be discharged, or how often it can be recharged.
Flow batteries are an entirely new approach to electric battery design. They have some similarities to fuel cells in the way they operate. In a conventional battery, the electrolyte chemicals are stationary within the cell. In a flow battery, the electrolyte is not stationary. Instead, there are four tanks. Electrolytes from two tanks with fresh ‘charged’ electrolytes are sent to the cell where they react chemically to produce electricity, while the spent electrolytes proceed to the two tanks for ‘spent’ electrolyte. During charging, the flow is reversed and electricity is applied to create fresh electrolyte. The total capacity of the battery is thus dependent only on the tank sizes (which can be scaled up to very large sizes). The throughput depends more than anything on the size of the cell’s reactive surfaces. The most common flow batteries so far are based on vanadium chemistry. Unfortunately this element is not very abundant. The specific energy and energy density are poor, in the range of 0.036 to 0.072 MJ/kg and 0.54 to 0.065 MJ/litre. Charge-discharge efficiency is in the range 75 to 80%. These cells are projected to be durable, lasting up to 20 years, and should be able to withstand upwards of 10,000 cycles.
Hydrogen from Electrolysis of Water
Today by far the most economical way to produce hydrogen is as a by-product of processing hydrocarbons and this accounts for over 95% of hydrogen production, but if the gas is to serve as an energy storage medium in a future world where hydrocarbons are scarce, then it must be obtainable from another source. The obvious answer to that problem is to obtain hydrogen through the electrolysis of water. Pure H2O is not a good conductor of electricity, so some ions need to be in the water to catalyze the process. In one common ‘hydrogen by electrolysis’ process, the hydrogen is a by-product of chlorine production from sea salt. This uses the reaction 2 NaCl + 2 H2O → Cl2 + H2 + 2 NaOH . A major manufacturer of electrolyzers, Norsk StatOil Hydro, uses potassium hydroxide (KOH) in the electrolyte instead of sea salt. Electrolyzer designs vary in detail. Norsk produces two types, one being “pressurized module electrolyzers” and a newer smaller-scale version uses proton exchange membranes (PEMs), so the latter one is much like a fuel cell in reverse. Another manufacturer of electrolyzers is Teledyne.
In the most simplified conceptual representation the electrolysis reaction takes the form 2 H2O -> 2H2 + O2. In other words, two water molecules can be dissociated into two molecules of hydrogen gas and one molecule of oxygen. This reaction is highly endothermic, i.e. it requires a large energy input. Electrolysis (using electricity to dissociate water molecules) is much more efficient than thermal decomposition, which requires very high temperatures. The resulting oxygen can be released while the hydrogen is stored. When hydrogen is burned, the reaction is reversed, a lot of energy is released, and the main waste product is water. Overall hydrogen is a very clean-burning fuel. In practice, there can be some oxides of nitrogen in the exhaust because like other fuels, hydrogen is usually burned in air which is mostly composed of N2.
As discussed in the post on energy storage for mobile applications, there are serious obstacles in the way of using hydrogen as a transportation fuel even though its performance as fuel is good per se . Notably there are difficulties with storing the material, but stationary applications relax many of the storage constraints. All three modalities of hydrogen storage are more plausible in a stationary setting than a mobile one. As a highly compressed gas, it can sit in a tank in an unmanned underground bunker. The tank can be instrumented to monitor it for risk of failure from hydrogen embrittlement, but if it ever does explode in most instances no one will be hurt. Storage using tanks with stabilizers (e.g. metal to form hydrides) may indeed require a stabilizer mass 50 times the mass of the stored H2, but in a stationary application the mass of the stabilizer doesn’t hurt too much although the cost may still bite. Finally, one might contemplate having large cryogenic tanks for the H2, in which case one would try to put to good use the H2 that will inevitably boil off.
As an electricity storage medium, hydrogen involves using electricity to electrolyze water, and then compressing and storing the hydrogen, and finally either burning it or reacting it in a fuel cell to produce electricity again. The electrolysis step alone has an efficiency in the range of 50 to 80%. Burning or reacting the hydrogen to produce electricity should have an efficiency of comparable to other combustion processes, also no better than 50%. This means that the overall efficiency of hydrogen energy storage is in the range of 25 to 40% at best, since we have assumed that compression and storage losses are zero.
Notes and Conclusions
The vast majority of the world’s dispatchable generation capacity today is in the form of fossil fuel burning thermal plants, and there isn’t enough untapped hydroelectric capacity to replace the thermal plants should they go out of service. If the dispatchable generation capacity provided today by fossil fuels is eventually replaced by bio hydrocarbons or re-formed hydrocarbons, then we will continue to store energy in large fuel tanks as we do now. However, if these new fuels prove inadequate, or if electricity produced by wind and photovoltaics claims a large share of future grids, then there is little doubt that on-grid energy storage will be necessary in that future. Making nuclear power plants more dispatchable is yet another approach to avoiding the need for storage. It’s an effort that France is pursuing.
There are a number of techniques that may work for storing electricity on-grid, but we have to consider their ecological sustainability, cost, and efficiency. Pumped water storage is probably the best idea on balance, but is only feasible in a select minority of locations. Compressed air storage has yet to prove itself. It might work out, but is less efficient and is also limited by the availability of good sites. Electric batteries are widely used today, but not on a utility scale. They are a very expensive way to go, and if utility-scale application of batteries ever does happen, they may be held back by the limited supply of certain materials such as vanadium. Recycling the key elements will be an important part of the industry. Hydrogen has the attraction that its supply in seawater is essentially unlimited, and the hydrogen fuel can be used for purposes other than generating electricity. But the electrolysis – to – electricity process suffers from poor efficiency.
The consequences of efficiency shortcomings should not be underestimated. For an electric utility, having a shortfall in generating capacity of even 5% is a major headache. The utility must buy from another utility over an interconnect, or cut off some customers, but in the long run it means increasing capacity, and that is costly. Now consider what storage efficiency means to the utility. In order to store a given amount of power with an efficiency of, say, 75%, the utility has to generate 33% more, and it has to have the capacity to do that. Adding a lot of capacity is not trivial – it’s a hugely expensive undertaking. Hydrogen, with an end-to-end efficiency of only 25 to 40%, leads to the need for primary generation of some three or four times as much power as is later released from storage. These enormous costs (which are on top of the cost of the storage system itself) are the reason why storage is used as a last resort. It is still far better to have adequate dispatchable generating capacity than to have on-grid storage, and that preference will probably persist forever even if cheap storage is eventually achieved, just because of the impossibility of ever making storage 100% efficient.