Concentrating Solar Power: Cost reduction potential

This article considers the economics of concentrated solar thermal plants, starting with a review of current costs and their sensitivities. Subsequent sections then look at the potential for reducing the costs of Concentrated Solar Power generation, and the consequences for cost competition with other technologies and when generating cost parity may be achieved.

Incorporation of solar thermal energy storage and/or auxiliary firing has an impact on the value of Concentrated Solar Power generation in electricity markets.

Today’s cost of CSP and its sensitivities

The structure of a commercial concentrated solar thermal project is very similar to other large power plant projects and typically involves several players.

An ‘Engineering, Procurement and Construction’ (EPC) contractor and its suppliers provide and warrant the technology to the owner, who finances it through equity investors, banks and eventually public grants. The owner gains revenues from the electricity off-taker (typically the electricity system operator) based on long-term power purchase agreements needed to pay off the debt and operation costs, and to generate a profit.

An operation and maintenance company provides services to the owner to operate the Concentrated Solar Power plant. This approach results in a complex contractual arrangement to distribute and manage the overall project risk, as the overall project cost of several hundred million euros typically cannot be backed by a single entity.

The perception and distribution of risks, as well as local and regional factors, strongly affect the cost, value and profitability of Concentrated Solar Power generation which depend on:

• the Engineering, Procurement and Construction price which, in turn, is dependent on technology choice, project size, country, site conditions, land costs, supplier’s structure, global prices for steel, etc.;

• annual operation and maintenance costs, determined by technology, size, site, availability of water, etc.;

• annual production of electricity, determined by technology, size, solar resource, and storage capacity;

• the rate paid for each kilowatt-hour (kWh) resulting from the political framework (in particular, the feed-in tariff and any capital subsidies) and the electricity market situation in the country;

• financing costs arising from the interest rate, project risk, technology risk, exchange rate, global economic situation, construction period; and

• project development costs, influenced by country specific factors such as the legal framework, currency exchange risks, tax and customs duties, etc.

Associated costs which are more difficult to quantify include impacts on rural landscapes, environmental taxes and abatement costs, specific charges on water or CO2 emissions, and, potentially, displacement of agriculture.

There is therefore no single figure for the costs of electricity from CSP, nor, for similar reasons, for other generating technologies to which it needs to be compared. One approach that is often used to compare costs of electricity generation is to calculate the ‘levelised electricity cost’ (LEC) which relates average annual capital and operating costs of the plant to the annual electricity production.

Recognising the limitations of the approach, particularly when comparing fossil-fired and renewable technologies where it does not capture differences in value to the customer, it nonetheless gives a useful ‘first cut’ view of comparative costs. For comparisons between fossil-fired plants and CSP with storage and/or supplementary firing, its limitations are less signifi cant as the technologies offer similar services.

Recent studies (IEA 2010b; Turchi 2010b; Kost and Schlegl 2010) give levelised costs of electricity from CSP of 15–22 € cents/kWh (20–29 US $ cents/kWh) in 2010 monetary values, depending on technology, size and solar resource.

To present an illustrative comparison of CSP electricity costs with other options, cost estimates for different technologies have been made taking data from a single source (US Department of Energy, 2010), and a simplified equation used to evaluate the LEC.

This analysis has assumed that the renewable energy systems (wind energy, photovoltaic (PV), and CSP) are positioned to have a favourable solar or wind resource and financing conditions.

For CSP a direct normal insolation (DNI) in Phoenix, Arizona (2500 kWh/m² per annum) is considered. The solar resource in Southern Europe is typically about 20% lower, whereas some sites in North Africa have a 5% higher resource potential. The impact on the cost is almost linear.

The analysis presented gives a cost figure for CSP electricity within the range given in the studies mentioned above (IEA 2010b; Turchi 2010b; Kost and Schlegl 2010). It also enables a comparison of the CSP generating cost to other conventional and renewable options under similar boundary conditions.

From the US Department of Energy study, it can be concluded that, when the solar resource is good, CSP had slightly lower costs than large-scale PV systems in 2010. (In 2011, the costs of PV systems were signifi cantly reduced so that they are currently slightly lower than those of CSP systems.)

CSP costs in 2010 were about twice those of onshore wind power farms, and slightly higher than estimates for offshore wind energy.

CSP can provide services similar to fossil fuel power plants in respect of dispatchable power and grid services as discussed later, but its electricity generation cost is today a factor 2–3 higher than for new fossil-fired power plants based on gas or coal. CO2 emissions of CSP plants are negligible compared with fossil-fired plants, and CSP would currently be cost competitive with coal if CO2 emissions were priced at about 80 to 120 €/t.

However, CO2 emissions certificates are currently traded in Europe at a rate of around 15 €/t, estimates of the social costs of carbon vary widely but are typically lower than this 80–120 €/t range, and there are other technical options that can avoid CO2 emissions at significantly lower costs than 80–120 €/t.

The implementation of CSP systems therefore currently depends on market incentives established by governments. However, changes in fuel prices, higher CO2 penalties and, in particular, cost reduction of CSP are expected to change this situation over time.

Cost reduction potential

Three main drivers for cost reduction are: scaling up, volume production and technology innovations. As an example, one of the first comprehensive studies of the potential for cost reduction of CSP was undertaken in the framework of the European ‘ECOSTAR’ project.

Scaling up

CSP technology favours big power plant configurations because:

• procurement of large amounts of solar field components can lead to discounts;

• engineering, planning and project development costs are essentially independent of the scale of the plant;

• operation and maintenance costs reduce with plant size; and

• large power blocks have higher efficiency than small ones and cost less per kilowatt.

The impact of scaling up on CSP electricity cost is still under discussion. The Kearney report (AT Kearney and ESTELA, 2010) indicates a 24% reduction of capital expenditure for an increase of parabolic trough plant size from 50 to 500 MW, and Lipman (2010) estimates a 30% reduction of LEC for an increase of turbine power from 50 MW to 250 MW. Finally, the Sargent and Lundy (2003) study points to a 14% cost reduction for a 400 MW power block.

Volume production

For parabolic trough plants, the Sargent and Lundy (2003) study estimates a cost reduction of 17% due to volume production effects when installing 600 MW per year. Cost decreases in the range 5–40%, depending on components, are expected in AT Kearney and ESTELA (2010).

Technology innovations

According to Pitz-Paal et al (2005), technology innovations will:

• increase power generation effi ciency, mainly through increasing operating temperature;

• reduce solar field costs by minimising component costs and optimising optical design; and

• reduce operational consumption of water and parasitic power.

Horizontal technological improvements are anticipated, potentially providing benefits across the families of CSP technologies. For mirrors these improvements include increasing reflectivity to 95% (by developing thinner front glass), anti-soiling and hydrophobic coatings on glass (to prevent dust deposition and reduce cleaning requirements), front surface aluminised reflectors, and polymer reflectors. Reflectance can be increased by 2.5% if the reflective surface is not covered by a glass layer.

This results in an increase in the collected power while the thermal losses that diminish it stay constant. The relative gain of the output power, which is the difference between collected power and heat losses, is about 3.5%. Replacing glass as a carrier of the reflective surface by other materials also offers a potential in a 25% cost reduction of the reflector.

Interrelated technology breakthroughs are expected in heat transfer fluids, storage media and thermodynamic cycles, as follows:

• Heat transfer fluids: superheated steam, new molten salts (with low melting temperature and higher working temperatures), nano-fluids, pressurised air (mainly development of new solar receivers), and circulating particles.

• Storage: phase change materials for direct steam generation, high-temperature storage for gas cycles, compact heat storage (chemical reactions), and heat transfer concepts.

• Thermodynamic cycles: supercritical steam or carbon dioxide cycles, air Brayton cycles and combined cycles (for tower technology).

To realise these technology breaktroughs and associated cost and efficiency improvements, it is essential to coordinate the different research, development and demonstration efforts with a market incentivation that favours cost reduction by innovation over cost reduction by mass production of state of the art technology options. Research without the chance to implement the technology in the market, and to improve and adapt it over a couple of technology generations, has a high risk of failure in a competitive market.

Increased research funding and a stronger integration of fundamental and applied research, together with demonstration programmes and market incentives, are required to speed up the innovation cycle. Fundamental research on new materials, heat transfer fluids, and coatings is needed, and integrated programmes should enable smooth progression of promising technologies from laboratory-scale prototype systems to pilot plants and demonstrations units.

Results of the individual phases should be independently evaluated and benchmarked with respect to their impact on system cost targets before starting on the next phase.

Competition with other technologies

In summary, it can be stated that different in-depth analyses of near- and mid-term technological options to reduce CSP costs have come to similar conclusions. They identify the potential for 25–35% reductions in CSP generating costs by capital cost and efficiency improvements based on technology developments already underway, and a further 20–30% reduction in costs through scaling up and volume production effects.

Operation and maintenance costs are also expected to decrease with CSP technology development and exploitation. For example, they dropped about 40%, from 4 $ cents/kWh (25% of the electricity cost in 1999) to 2.5 $ cents/kWh, at the Kramer Junction plant in the US between 1992 and 1998 (Cohen et al., 1999). Operation and maintenance costs also reduce sharply as plant size increases.

To estimate whether the anticipated cost reduction may enable CSP to break even with the LEC of the fossil-fired alternatives plots the
percentage changes in investment cost required for CSP plants to break even with coal- and gas-fired plants as a function of fuel price. At today’s fuel prices, a reduction of 50–70% in the investment cost of CSP is needed to compete.

Prices of CO2 certificates will influence the point at which cost competitiveness is achieved as they can be considered as equivalent to surcharges on the fuel price. For coal, each additional euro per tonne CO2 on the certificate price has a similar effect on the competitiveness as a CSP cost reduction of 0.5% (for gas it is 0.3%).

Assuming, for example, a coal price of 15 €/MWh and a CO2 certificate price of 30 €/ton in the future, a 30% cost reduction of the CSP plant corresponds to break-even on LEC with mid-load coalfired plants. The analyses discussed above consistently point to the potential for signifi cantly greater CSP cost reductions.

The competition is also strongly determined by the cost of money as the cost per megawatt of capacity of CSP systems is larger than that of fossil fuel fired power plants. The overall global market situation, as well as the perceived risk of the investment, strongly influence the cost of money for a project.

However, typically the loan conditions are known and fixed at the beginning for the pay-back time of the project, whereas fossil fuel price change represents a continuous risk.

The competition with other renewable technologies, in particular with solar energy PV (including concentrating solar PV) which uses the same solar resource, is more complex.

Decentralised application of solar PV competes at the level of consumer prices, which are significantly higher than the market prices for bulk electricity. In Europe, grid parity for domestic solar PV systems is expected to be achieved within the next few years. The market growth of solar PV in this segment will reduce the amount of electricity taken from the grid, but will force the grid to react more quickly to the changes provided by this variable resource.

The flexibility of CSP can be one option to help the grid accommodate such variable sources. If solar PV is used to provide bulk electricity, its average value is lower than CSP as it cannot provide dispatchable electricity, and cannot provide other grid services (stable frequency, spinning reserve, etc.).

On the other hand, the cost-reduction curve for PV has to date been very steep, the PV market and PV research capacity are currently much larger than for CSP, and PV power plants can be implemented more quickly than CSP systems.

Recent aggressive competition, in particular from Asia, has resulted in a further price drop of PV systems and has led to a situation where in some markets, where time of delivery and capacity aspects are not refl ected in the revenues, project developers have preferred large-scale PV over CSP technology options.

However, the potential future cost reductions of both CSP and PV are high, and only time will tell which will have the steeper learning curve.

The difference in value between the technologies depends on the overall energy system and, in particular, on the share of variable renewable electricity, and hence needs to be evaluated for each market.

The future cost evolution of solar PV and CSP systems, and the price difference between dispatchable and non-dispatchable electricity, will be decisive in determining the relative sizes of the contributions of solar PV and CSP in the market.

Given the challenge that society faces in transforming quickly to a low carbon economy, and taking into account the high resource potential that solar energy has in the world, it would be inappropriate to drop one or the other option too early based on short-term price differences.

CSP’s ability to support the system integration of variable renewable sources, also suggests that its further support should not be determined solely by its short-term competitiveness with PV systems.

Time-frames for cost competitiveness

An alternative approach to estimating future potential for cost reduction is to use well-established ‘learning curve’ effects, which are based on observations for technologies more generally that their cost reduces by a characteristic percentage for each doubling of installed capacity (hence, the ‘learning rate’ is defi ned as the percentage reduction in costs for each doubling of installed capacity).

Although this concept was originally applied to a product of a single entrepreneurial entity it has been found to work for many mass produced components on the global scale.

If the concept is applied to a system that consists of different components like a CSP plant, the overall learning curve for the system will be, at least in part, an amalgamation of the learning curves of individual components.

While solar collectors or thermal storage systems do not yet have the status of being mass-produced, the conventional power block is. Further implementation of solar power plants will therefore only marginally impact its general future cost reduction, although there may be potential cost reduction for CSP associated with its adaptation to the specifi c needs of CSP applications.

Trieb has suggested an approach that combines different learning rates of components and the effects of scaling to larger plants for CSP, and calculated a CSP system learning rate of 14%.

The uncertainty in this figure is high as it is not based on empirical data. The following analysis, which examines cost reductions up to 50%, therefore considers a range of 10–20% as potentially achievable for CSP. The impact of installed capacity on costs for this range of learning rates.

Starting from an actual installed capacity of 1 GW, a 20% learning rate would require an installed capacity of around 9 GW to halve costs, whereas 100 GW would be required in the case of a 10% learning rate.

Starting from a current CSP installation rate of around 500 MW per year, and assuming a growth rate in CSP installations of 15% (low) and 30% (high) per year, results in CSP achieving a 50% cost reduction between 2021 and 2031.

The learning rate and the growth rate of installed CSP capacity are key determinants of when CSP will be cost competitive with other technologies. The ranges of figures selected in this analysis are based on expert estimates and opinion, and have not been verified by actual data (which are not available).

It is therefore strongly recommended that mechanisms are put in place that enforce a transparent monitoring of installation costs, and the rate of CSP technology capacity increases, to enable estimates of the learning rate to be refined.

The growth rate of the CSP market is currently constrained by market opportunities rather than production capacity. Additional incentives, and the creation of new market opportunities in other countries, will help to speed up the cost reduction process according to this model.