Jonathan Howes, Isentropic, UK
|Jonathan Howes, technical director, Isentropic|
The two primary drivers for the use of renewable energy sources are the need for sustainability of energy supplies and the climate change-driven desire to decarbonize power generation. A secondary, but nonetheless important concern, is the independence of supply to protect against the use of energy as a political weapon.
The Need for Flexibility
Classic renewable energy sources are not the only solutions to these problems, nuclear power being an obvious example. Concerns over the environmental implications regarding the disposal of waste are largely emotional rather than objective since, by comparison with fossil fuel based generation, current nuclear technologies are exceptionally clean.
Both renewable – with the possible exception of geothermal energy – and nuclear options, however, suffer from a similar Achilles heel as nuclear power stations do not operate well at widely varying loads and the cycles found in renewable sources rarely synchronize with demand. The inflexibility of nuclear power, by comparison, results in a need for over-capacity if the peaks in demand are to be accommodated.
The ability to provide a load-levelling buffer for renewable sources, or a means to absorb otherwise redundant generation for later use in the case of nuclear, must therefore be seen as highly desirable.
In addition to the limitations of current clean, also known as low-carbon, generation technologies, electrical distribution networks are also capacity limited. A sudden increase in local demand can result in system failures within the grid. The ability to buffer the energy supply locally would allow the storage node to be slowly replenished via a limited grid connection and then released into the local distribution network, without overloading the primary grid connection.
A recent example of local grid overload occurred in the UK in the county of Cumbria where, after the loss of gas supplies due to extensive flooding, electric heaters were provided to local residents as an interim measure. During the accompanying cold snap, many of these heaters were switched on at the same time, resulting in a fire at a substation. If, as seems probable, gas heating is progressively replaced by heat pumps, problems such as this are likely to occur on a much greater scale in the future.
A typical UK demand cycle shows a minimum at around midnight, with a morning surge in demand followed by a plateau until the end of the working day. A peak occurs in the early evening followed by a progressive reduction to the minimum. As an example of the variability of renewable sources, a typical 24-hour wind farm output cycle is shown in Figure 1. This bears absolutely no relationship to the demand cycle and, without storage of some form, the peaks are liable to be wasted, while the troughs will not provide a useful contribution.
|Figure 1: Typical wind farm output over a 24-hour period|
It is common to find an excess of power generated when prices are at their lowest and a low output at periods of high demand. This severely impacts the ability of wind farms to compete economically with each other. More traditional forms of generation and the widespread adoption of wind energy have so far been driven primarily by subsidy rather than by sound engineering and economic decision-making. The problem with subsidy systems to encourage the adoption of a new technology is that methods to extract the subsidy become more attractive to investors than the desire to solve the primary problem that led to the subsidy in the first instance.
|Schematic of a 2 MW, 16 MWh pumped heat electrical storage plant designed by Isentropic|
The desirability of wind power is further damaged by the difficulties that this extreme variability of wind farm output imposes on grid synchronization and control. Again, the addition of a storage technology acting as an intermediary between the farm and the grid allows proper regulation of phase and frequency, and will significantly increase the value of this form of power generation.
Other renewable sources, while not suffering the extreme variability of wind power, also have related problems. Tidal power in the UK has four peaks per daily cycle on an approximate 24-hour and 50-minute cycle. This rarely aligns with the demand cycle. Wave energy is, of course, dependent on wind but, due to the inertia inherent in ocean wave systems, does not exhibit the extreme short-term variability of wind power.
Nuclear power sources, not being amenable to operation at significantly varying power levels, lead to the handling of variations in demand by providing an excess of generation capacity to handle peak loads and then simply dumping this excess when it is not required. This is both wasteful of equipment and an inefficient use of energy. Implementation of significant amounts of storage capacity could, therefore, reduce the need for primary generation capacity with significant cost savings, if a suitable technology can be identified.
CURRENT STORAGE TECHNOLOGY INADEQUATE
Current solutions to the demand and supply mismatch comprise pumped hydro storage and gas turbine plants, both of which can be brought on-stream rapidly. Gas turbines are incompatible with the objectives of a low-carbon economy and pumped hydro, although an excellent and highly developed technology, is dependent on suitable geography. Planning and environmental constraints, as well as the limited number of suitable sites, make significant expansion of this method of storage very difficult.
Escovale, a specialist energy storage consultancy, has estimated that there is currently a 200 GW latent opportunity for a technology that can match pumped hydro on cost, but is geographically independent. This is estimated to be equivalent to a $200 billion market. Most currently emerging storage technologies are nowhere close to this and a typical target is $1000 per kW/$100 per kWh.
As a guide to the relative costs of some of the various storage systems, Table 1 provides typical values for cost per kW and cost per kWh. Due to the seasonal variation in demand and supply there has been some interest in seasonal versus short-term storage. This is problematic since a seasonal store represents a latent cost, the cost of storage per unit energy reducing sharply as the number of charge-discharge cycles is increased.
This calculation is also adversely affected by the relative effectiveness of small versus large amounts of storage. As an illustration of this, it is informative to review some typical wind farm output data in conjunction with an energy storage model.
For the purposes of this exercise, a round trip efficiency (RTE) of 75 per cent has been assumed. The storage capacity is expressed in terms of days of average wind farm output, with the farm in the example delivering an average utilization of 33 per cent of peak power. A simplified store utilization model has been applied in which the following rules apply:
- A constant target baseload power output is specified as a percentage of average output;
- If direct wind power falls below the specified baseload the store is depleted;
- If the direct wind power exceeds the baseload the store is charged, but energy transferred to the store is penalized by the RTE;
- If the store is fully charged any excess power may be output, such that the baseload level is exceeded.
Many other more realistic control laws are possible, however, this serves to illustrate the impact of the addition of simulated storage to real data.
As illustrated in Figure 1, wind farm output is extremely variable ranging from a peak of around three times the mean power output to nothing, sometimes for several days. Setting a target of complete reliability of baseload delivery and plotting the maximum baseload that can be sustained on this basis against the installed amount of storage results in Figure 2.
|Figure 2: Percentage of mean wind farm power for a fully reliable supply with increasing amounts of storage capacity|
The steepness of this curve at the lower levels of storage shows that the improvement in reliability per unit storage is extremely high. A law of diminishing returns is evident as the storage capacity is increased and it is clear that a target baseload of 90 per cent of mean power output is close to the maximum possible limit. This limit reflects the impact of the RTE, since some energy becomes unavailable as a result of losses within the storage system.
Setting the store capacity to seven days and setting the baseload such that baseload failures are just avoided results in Figure 3. This shows the store energy content as a percentage of the installed value and the resulting power output assuming that any excess energy is output.
|Figure 3: Store utilization and output – wind farm with a storage capcity of seven days and a baseload of 50.6 %|
The result is still extremely noisy because of power spikes occurring where excess power is available but the store is already fully charged. However, the wind farm owner would be in a position to guarantee a certain level of power and frequency reliability, with a probable increase in the value of the electricity supplied, although the excess generation would be attended with exactly the same problems as any other wind farm output.
Due to the innate cost of latent storage, that is, storage that is not being frequently cycled and the relative improvement in load-smoothing for smaller amounts of storage, it is unlikely that current technologies can satisfy long-term storage needs. The market is likely to change as renewables become more widespread, leading to an increased value added by storage. It would be a serious error to size storage projects based only on a current market model.
ROLE OF ELECTRIC VEHICLES
Other forms of implicit storage will also have some impact. Electric vehicles have received some attention in this regard and battery swap-out concepts, rather than charge in-situ, may result in a need for significant amounts of energy during off-peak times.
The concept of the ‘smart grid’, where control of some distributed loads, for example, refrigeration and storage heating, is passed to the power supplier rather than the consumer, has also been proposed as a load-levelling device. The addition of distributed storage would achieve the same effect, many smaller storage plants then acting as system dampers without the need for complex control systems. This has the further benefit of having no impact on consumer behaviour and hence distributed storage appears to be a significantly better option.
A hydrogen economy is also of interest here as a potential consumer of off-peak power. Although the RTE of hydrogen is poor, it does have the benefit of being a high-grade fuel suitable for high energy density applications, such as transport, including aviation. Given concerns over the acidification of the world’s oceans, a side benefit may be that the electrolysis of seawater can produce strong alkalis as a by-product.
Notwithstanding implicit storage technologies, there remains a need to load-level throughout a grid system that is likely to experience increased utilization as a result of technological advances, such as the widespread use of heat pumping. Modulation of intermittent renewables before feed-in to the grid is also a clear and developing need and so the future picture will almost certainly comprise some blend of both direct and implicit storage. The future for new storage technologies looks very bright indeed.
Jonathan Howes is technical director of Isentropic, a UK-based private technology development company, which aims to change the way the world approaches energy generation and storage.