Thermal energy storage can be beneficial for integrating Concentrated Solar Power into an electricity system in both these spheres. Inclusion of a storage system in a solar thermal plant can therefore have a significant impact on its value, which comprises three main components:
• the value of the kilowatt-hours of electrical energy generated by the solar thermal plant, which will vary over time in a competitive electricity market, reflecting the availability and cost of electricity from other sources;
• the contribution that the Concentrating Solar Power plant makes to ensuring that generating capacity is available to meet peak electricity system demand; and
• the ‘services’ provided by the plant in helping the electricity transmission system operator to balance supply and demand in the short term (typically, on timescales of seconds and minutes).
Considering the first component of value, optimisation of the relative sizes of a Concentrating Thermal Solar Power plant’s collector field, turbine and thermal energy store will depend crucially on the structure of the price curve (the hourly variation of electricity price through the year), which in turn depends on the supply demand-pattern of the electricity system into which the CSP plant feeds.
Generally, the value of a kilowatt-hour of electrical energy is higher at times of higher demand. Even without storage, the profile of output from a CSP plant in Southern Europe and the MENA region is reasonably well matched to demand, which often peaks in the middle of the day when the sun’s strength, and hence CSP generation, is highest.
Demand often remains strong into the evening, and storage enables some proportion of the daily generating capacity of the CSP plant to be shifted to the evening to contribute to meeting this demand and so enabling the CSP plant to benefit from the associated revenues.
The ability of a CSP plant with storage to match the pattern of diurnal demand has been well received by the power grid operator in Spain, Red Eléctrica de España (REE). This demand pattern is typical for Europe more generally, with electricity prices generally peaking at midday and in the early evening, although this varies between week-days and week-ends, by season, and by country.
Although renewable systems without storage or back-up firing will be able to match the demand curve statistically quite well, there is still a need for ‘shadow’ capacity to ensure security of supply. CSP systems (in particular when equipped with fossil co-firing) can avoid this need.
Given the coincidence of the energy generated by CSP (as by any other solar energy technology) and the price peaks in the middle of the day, storing solar energy as thermal energy rather than supplying electricity to the grid immediately at the time of solar irradiation would regularly be associated with an opportunity cost at the system level.
Energy losses associated with storing and retrieving heat exacerbate this opportunity cost although, in practice, the large volume to surface ratio of the storage containers and their good insulation means that energy losses are low. ‘Round trip effi ciencies’ of 93% have been routinely achieved by commercial plants in Spain, even when energy is stored for 24 hours.
Scale-up of plant sizes should further reduce heat losses as the surface area of storage tanks will reduce in relation to the stored volume. The economic value of thermal energy storage for a CSP plant cannot therefore be calculated at the plant level, but only at the system level: the overall configuration of the electricity system determines the price curve and hence the value of shifting the timing of generation through the day.
Generally speaking, the higher the share of solar power within the system, the less pronounced the diurnal price curve will be, reflecting a need to use solar power at times other than the middle of the day peak. This implies that thermal energy storage is less relevant today (at low solar shares), but may rise over time (with increasing solar shares).
A recent simulation by the Institute of Energy Economics at the University of Cologne confirms this effect. It involved a least cost optimisation for the (stylised) development of the power markets of the Iberian Peninsula (i.e. Spain and Portugal) until 2050. Allowing a choice between CSP systems with different thermal energy storage sizes, the model indicated that the cost optimal solution only involves signifi cant amounts of CSP with thermal energy storage in the long-term.
In the short to medium-term, electricity prices in the model are sufficiently high during the day (and low at night) that it is best for CSP plants to sell electricity as it is generated. In the longer term, the model includes substantial capacities of variable renewable energy sources, especially PV and CSP without storage. This has the effect of lowering, and in some cases reversing, the differential in electricity prices between day and night, making it economic to include thermal energy storage in CSP plants so that they can take advantage of the better prices when the sun is not shining.
Two key insights into the value of thermal energy storage in CSP plants in Europe emerge from this simulation exercise:
• The opportunity cost of thermal energy storage at the system level (i.e. the cost of transferring electricity from a time of high prices to a time of lower prices) can in fact exceed the benefits of thermal energy storage at the plant level.
• Whether or not this is the case largely depends on the share of variable renewable energy supplies in the overall electricity system. Depending on the overall configuration of an electricity system (i.e. the mix of power plants, availability of pumped storage, demand level and demand structure), the amount of variable renewable energy supplies has to reach a specifi c threshold before the price differences between hours with high solar radiation and hours with low or no solar radiation decline or even reverse.
On the issue of seasonal patterns of electricity supply and demand, CSP storage is unable to overcome potential variations in the price curve which might arise from seasonal patterns of generation by renewable sources. For example, CSP plant generation in Southern Europe on a typical sunny day in winter will only be around half that on a sunny day in the summer. Again, the appropriate response will depend on the properties of the electricity system overall, i.e. on the seasonal pattern of demand and the other sources of generation in the system.
It is noted that seasonal fluctuations of electricity from CSP plants located in the MENA region are lower than those in Europe, and hence for Europe, importing CSP electricity from MENA countries may be able to make some contribution to addressing seasonality.
Other storage technologies (besides pumped-hydro), i.e. ‘unconventional’ storage systems (e.g. compressed air energy storage), are not cost-efficient for use as seasonal storage, because the investment is only used for a limited time during the year and not every day, even taking into account the planned expansion of variable renewables.
However, like CSP they may find application in daily and weekly electricity storage. In practice, the combination of regional variation in renewable energy supplies, fossil back-up generation, and sufficient grid interconnection typically prevents the prolonged, substantial price peaks which would be required to make such ‘unconventional’ seasonal storage systems cost efficient from the system perspective.
This is especially so when other potential options for added flexibility are taken into account such as the use of biomass, conversion of electricity to gas and the use of the gas grid, or the use of demand side options especially in the industrial sector.
However, in the absence of a well integrated electricity system, supplementary fi ring with natural gas or biomass may, in some circumstances, have value for the electricity system in helping to balance seasonal variations in generation and demand. It should be noted, however, that the use of local biomass for this purpose would rely on a good annual rainfall (to enable the growth of the plants and trees providing the biomass) combined with high direct solar radiation.
With regard to the second component of value, the provision of generating capacity to meet peak electricity system demand, CSP with storage can contribute to meeting peak system loads and can provide backup capacity to cover variable renewable sources.
Incorporation of supplementary firing will further increase the capability of the CSP plant to provide capacity at the system peak, although the efficiency of fossil-fuel use for such supplementary firing is likely to be signifi cantly lower than if it is used in a combined cycle power plant. The value of providing capacity to meet the system peak demand will depend on the system, so its quantification needs to be informed by system models.
The electricity system operator needs to know the profi le of electricity generation it can expect from its connected plants over the next day or two. Although further improvements could be made, weather forecasts are already suffi ciently good that the output from CSP plants over such time periods can be predicted with high confi dence.
For example, in Spain, CSP plant operators must predict their electricity production 24 hours in advance with a maximum deviation of 10%, and 6 hours in advance with only a 5% deviation. These tight requirements imposed by the Spanish grid operator, REE, are regularly fulfi lled by operational CSP plants. In contrast, deviations greater than 25% are usual in the predictions made by the Spanish wind farms.
Turning to the third component, the value of thermal energy storage in enabling the CSP plant to deliver grid services, such services may be differentiated according to response timescales: ‘regulation’ services requiring response time measured in seconds, ‘spinning reserves’ being available on timescales of up to 30–60 minutes, and ‘non-spinning reserves’ capable of being started up and brought on line within 30–60 minutes.
A CSP plant, with or without storage, is considered to be unlikely to make a significant contribution to regulation or non-spinning reserves services. In the case of regulation services, this is because the inherent storage in the steam generator is small (in conventional plants it is the steam drum which is the initial source for energy ramps on timescales of seconds), and the inertia of other plant components prevents a suffi ciently fast response. In the case of the longer-term non-spinning reserves, this is because either the CSP plant will be running and delivering electricity, and not kept in reserve, or if shut-down may not be able to be started up quickly enough, though this depends on the specific technology.
CSP with storage can provide spinning reserves, being able to ramp up power if operating at part-load in less than 30 minutes by drawing on the stored heat (the rate of ramping is limited by the thermal inertia of the equipment). Ramping down is quicker: on timescales of around 15 minutes by diverting heat to storage. This is used in Spain to deliver, on demand, 30% power ramps in less than an hour, enabling the plant to be considered dispatchable by the grid operator REE.
In discussing the services provided by a CSP plant in helping the electricity system operator to address short-term supply demand imbalances, consideration must be given to the potential ‘negative value’ arising from transients during partly cloudy days. Inclusion of
at least three hours storage in the CSP plant enables the substantial thermal inertia provided by the storage medium to be used to dampen any resulting steam temperature/pressure gradients at the power block inlet on such days.
CSP plants may also be able to contribute to grid services by providing ‘reactive power’ which is needed to achieve local balance on the system. Small payments are made to CSP plant operators in Spain for supplying reactive power. However, CSP plants located remote from demands are unlikely to be able to make a substantial contribution to meeting system operational needs for reactive power.
Whether or not thermal energy storage is the cheapest and/or simplest way of delivering such grid services merits further investigation. It can be assumed, however, that the value of grid services provided by thermal energy storage increases as the concentration of solar power plants (CSP and PV) in a particular region increases.
Sioshansi and Denholm (2010) have undertaken system modelling studies to evaluate incorporation of thermal energy storage in CSP plants in four locations in the southwest USA, which confirm the system dependence of the value of storage discussed above. In all four cases, for the modelling assumptions used in the study, reductions in the cost of storage are needed to make storage economic if just the energy value of kilowatt-hours sold is considered. However, in this study, inclusion of calculated values of providing system services and capacity substantially increases the value of storage, making it economic in all but one of the 16 site and parameter variations considered.
The value of auxiliary firing
A CSP plant with storage and auxiliary firing can reproduce many of the operating characteristics of fossil-fired plants or dispatchable hydro plants (the match becoming closer as the auxiliary firing capacity of the CSP plant is increased). In this way it is more readily integrated into normal power system operations than other renewable electricity sources such as wind or PV. Its output can be scheduled to suit its host power system, or where plant scheduling is based on a market, to run during the time of day when prices are highest.
Wind energy and PV can be linked with pumped storage hydro to deliver some of these benefits, but the round trip losses are very much greater than the losses in thermal storage associated with CSP. Supplementary firing, if installed, can also be used to smooth power block operation on cloudy days, and thereby deliver system services.
Again, however, the economic value of auxiliary firing at the CSP plant level has to be carefully compared with the alternative options within the power system, for example, high-effi ciency thermal power plants located closer to the demand centres. Most of today’s CSP plants operate at signifi cantly lower fuel-to-electricity efficiencies than conventional power plants, so auxiliary firing can have a negative impact on CO2 emissions. However special designs optimised for hybrid operation can go some way towards overcoming this problem.
The value of supplementary firing (and storage) will tend to be higher when the accessible electricity system is smaller. For many countries in the MENA region, the size of the electricity system into which power plants feed is limited. This factor, and the cost advantages of natural gas firing, may go some way to explaining why supplementary firing and hybrid schemes (in which a CSP facility is used to augment the effi ciency of a larger fossil-fired plant), have previously been adopted in this region. As fossil-fuel costs increase there may be a shift to thermal storage as the preferred mechanism for addressing the isolation issue.
Also, CSP may be deployed along with other renewable technologies such as wind power and solar PV which may contribute to increasing the reliability of supply in the absence of good grid connections.
When auxiliary firing is incorporated, a CSP plant may be able to help the system operator in a ‘black start’ situation, i.e. to supply electricity to the system when it is not energised in order to restore electricity supplies. This capability would have to be designed into the plant.