The role of flywheel energy storage in decarbonised electrical power systems

Flywheel technology has the potential to be a key part of our Energy Storage needs, writes Prof. Keith Robert Pullen:

Electricity power systems are going through a major transition away from centralised fossil and nuclear based generation towards renewables, driven mainly by substantial cost reductions in solar PV and wind. This transition, accelerated by government subsidies, has reached a self-sustaining momentum which will only accelerate as markets are reformed to be more consistent with this new order. Intermittency in demand has always been present short duration events were balanced passively by virtue of the rotating inertia of steam turbines and generators of large fossil and nuclear stations. Anything more than 10s of seconds required starting or peaking stations and/or pumped hydro storage. With the replacement of large stations, the supply is now intermittent and the stabilising inherent inertia is steadily being removed.

It is vital that the frequency of the AC supply is kept within around ±1% of mean otherwise the system collapses. The issue so far has been dealt with by a combination of demand side management and storage, the latter mainly using large banks of Lithium-Ion (Li-Ion) batteries. Li-Ion can meet the sub second demand with appropriate control electronics and can also provide durations of typically one to two hours allowing additional revenue streams such as load shifting, arbitrage and providing other grid ancillary services. An additional factor has been the specification of long periods of duration in ancillary contracts. This has prevented high power, low storage technologies such as flywheels and supercapacitors competing with Li-Ion in auctions, such as the UK’s Enhanced Frequency response requiring a duration of 15 mins. The physics of the grid stability problem does not require such a long timescale.

The question is whether technologies other than Li-Ion may be better placed to meet the growing future storage needs rather than deploying greater numbers of grid scale Li-Ion batteries. This question is not simple to answer since there are other changes taking place which will greatly affect the future with great interdependency. Taking first the technical front, the predicted mass take up of electric vehicles will have a substantial effect on grid balancing. If done well, this could help reduce the balancing problem if charging can be delayed to periods of low demand, denoted smart charging. Taking this a step further, the storage in the battery in plugged in vehicles could absorb or provide power into the grid in a Vehicle-to-Grid (V2G) scenario. However, this might be limited if the battery is already charged or drawing power compromises the life of the battery. There are already concerns about the lack of recycling of Li-Ion batteries and their reliance on materials from questionable sources, so anything which reduces vehicle battery life is of concern.

Another technical innovation is demand side response, whereby demand can be controlled but delaying power draw for equipment which has a slow time constant. For example, heater-chiller units normally switching according to a thermostat could be delayed or started earlier. This is already being done and there may be other ways to time shift demand. Another technical effect is the expansion of interconnectors between countries. Finally, the market structure is in the process of being reformed from one originally devised for a system of large centralised power stations and peaking plants to one fit for purpose for a system with high penetration of renewables, distributed generation and prosumers. Some argue that if market reform is done well, the best and most efficient technology will emerge without need for intervention.

How does this affect futures needs for storage?  The consensus is that all options should be kept open but more specifically there is a need for storage across three timescales:

• fast response, high cycle storage,
• bulk storage with at least 8 hours rather than the 1-2 offered by Li-Ion
• > 1 day to seasonal storage

There is an understanding that storage for more than a day cannot be met by anything other than chemical with natural gas being a necessary evil until this can be replaced with renewable fuels, likely hydrogen, but also ammonia or non-fossil derived hydrocarbons.  

For fast response and the 8 hours need, one could argue that lowering the C rating (ratio of kW/kWh) would allow Li-Ion to meet this, but it increases cost. Added to that there is a desire to reduce energy storage costs further and also employ technologies that have lifetimes of over 20 years with low CO₂ in manufacture, which are easily recyclable unlike Li-Ion. Better candidates include compressed or liquid air, flow batteries, gravity systems, pumped hydro and engines running on renewable fuels. However, none of these can meet sub second response from start up, so there is a gap to be filled. It makes less sense to use Li-Ion to meet this gap since there would be an overlap of provision in duration and Li-Ion suffers from limited cycle life. The best choice is the lowest cost technology with low minutes of storage and flywheels fit this perfectly.

A flywheel is a very simple device, storing energy in rotational momentum which can be operated as an electrical storage by incorporating a direct drive motor-generator (M/G) as shown in Figure 1. The electrical power to and from the M/G is transferred to the grid via inverter power electronics in a similar way to a battery or any other non-synchronous device. In order to keep the size of the M/G reasonable, the flywheel is operated between a minimum and maximum speed and would be kept spinning by means of a small input power to make up for the parasitic losses. The minimum speed of the flywheel is typically half its full speed, the storage energy is be given by ½ (1²-0.5²) I_fw_f² where I_f is the rotor moment of inertia in kgm² and the w_f maximum rotational speed in rad/s. The power level is controlled by the size of the M/G, so this is independent of the rotor. In this way, very high powers are possible with a relatively small flywheel sufficient for a few 10s of seconds or minutes. Parasitic losses occurring due to aerodynamic drag or windage can be almost eliminated by use of a sealed casing which holds the required vacuum. Bearings create a drag loss but can be minimised by levitating the rotor using a permanent magnet passive magnetic axial bearing. Low loss mechanical bearings or active magnetic bearing may be used as radial bearing to keep the rotor stable. A further loss may be developed in the M/G but this can be reduced by careful design choices. Depending on the design particulars, the input power to sustain the rotor at speed can be low and similar to auxiliaries needed to cool or heat batteries. Given their robustness, flywheels typically can operate at any ambient temperature with some air or water cooling of the M/G and power electronics when power is drawn.

The key technology in any flywheel is the rotor. Initially flywheels were made of solid metallic steel either run at low enough speed to ensure burst would never occur or substantial containment was provided in the case that higher speeds were used. There are examples of these designs in operation today with much recent activity. There was a flurry of interest in use of carbon fibre composites for flywheels starting from the 1980s. This material has a specific strength (strength/density) of around 10 times that of steel in direct tension so it appeared to be vastly superior as a flywheel material offering a breakthrough.

However, since the material has its strength in one direction and material properties vary, carbon composite flywheels more typically have a rotor weight of around 3-4 times less than solid steel for the same storage but the overall volume of the rotor is around twice. Still a substantial advantage for transport applications but at higher cost. More recently it was found that composite flywheels can fail in an explosive mode which created very high pressures within the containment. This can be contained albeit with very thick containment, composite containment or by burying the flywheel in an underground bunker. This issue severely detracts from the lightweight rotor when the whole system weight and size are considered. Another approach is to laminate a steel rotor such that in the event of a failure, only a small part of the rotor is released. This allows the casing to be thinner and yet maintain safety so such flywheels can be above ground or in buildings without additional safety measures. Steel is fully recyclable so even after decade of operation, use of this material is highly sustainable.

Although any individual flywheel may be small relatively to grid requirements, any power level can be achieved for flywheels since modules are simply multiplexed in the same way as grid scale battery consists of many cells and strings. Figure 2 shows a layout of an 8MW array that can be fitted inside a 40 foot container as an example.

More information on flywheel applications can be found in:

Amiryar M. and Pullen K. R., “A Review of Flywheel Energy Storage System Technologies and Their Applications”, Journal of Applied Sciences-Basal 7(3), Article number ARTN 286, Mar 2017