4. Economics of Solar Photovoltaic Generation

At the end of the last decade, approximately 10 kg of silicon was required in order to produce 1 kW of PV cells (or 6.66 m² of modules). In 2013, this magnitude was 5.6 kg/kW. The wafers were thinner, and thus, their number per ingot was higher and the quantity of material lost in the cutting was lower. Of course, the thinner the wafers, the more it is breakable, which requires a more careful processing.

The entry barriers in the PV value chain are decreasing in the downstream direction. They are very high regarding the manufacturing of silicon and almost null in the case of the promotion and installation of PV farms.

The greater p-Si market is the long-term one. In this case, the prices only reached $75–$90/kg.

It should be indicated that there was a widespread belief that the production cost was around $35–$40/kg. The pressure of demand for PV would have considerably increased the profits of the manufacturers, and thus, it would have facilitated the financing of new investments in production capacity. After such increase, the manufacturing cost would have fallen by half.

It should be warned that the data in Table 4.1 are not as homogenous as it would be desirable. Unfortunately, some firms do not clearly distinguish between installed production capacity, production and sales of cells and/or modules in the year in question. In reality, only the last figure is the one in the table. Moreover, data on the sales of several solar PV companies are not publicly available. This information is currently in the hands of private consultancies which sell it at prices which are very high for independent researchers. The most important free sources are those cited in the text.

Although there have been statements on new entrants as well as on capacity expansions of already existing firms, the credibility of these statements is limited. The plans are too often exaggerated, and the execution deadlines are not met. Furthermore, when the end of construction of a new plant is announced, only the installation of the equipment is often completed, but the manufacturing lines are not operative. Such an announcement does not make much sense sometimes since the plant is empty (REN21 2011: 41). Therefore, there is a huge divergence between the announced production capacities, the real production volumes and the sale of products (wafers, cells or modules) realized. For example, Drury et al. (2012: 10–14) publish the production capacities of four important module manufacturing firms (First Solar, Wuxi Suntech, Yingli Green Energy and Trina Solar) between 2005 and 2011. Comparing those capacities with the sales in Table 4.1, the excess of average capacities is about 34 %. According to the published data by Photon (www.​photon.​info, 15 September 2015), the utilization rates of module manufacturing facilities worldwide reached their peak in 2011, when the global production capacity was 52 GW and the total manufacturing output reached 36.6 GW. This involves a utilization rate of 70 %. According to this source, this rate was 66 % in 2013.

Since the criterion in order to be listed in the table is a remarkable presence in the last year being considered, some relevant firms in the first years are missing. Their production volumes, however, were very low with respect to those which are now common. In reality, those firms were unable to gain market quota. They were taken over by another or have disappeared.

An Original Equipment Manufacturer (OEM) is a company which produces a subsystem or component which is to be assembled in an end product of another firm.

According to the data published at the end of 2014 by the Photon International agency, Trina Solar (with 3.61–3.66 GW), Yingli Green Energy (3.6–3.8 GW) and Canadian Solar (2.73–2.78) have been the three largest module suppliers in 2014. Moreover, Jinko Solar moved up to fourth place and JA Solar climbed to fifth place. Meanwhile, Sharp Solar dropped two spots to sixth. Rounding out the top 10 were ReneSola, First Solar, Hanwha SolarOne and SunPower and Kyocera tied for tenth place. The predictions for 2015 indicate a clear expansion of the sales: from 4.9 to 5.1 GW by Trina Solar, Canadian Solar expects to reach between 4 and 4.3 GW, Yingli G.E. 3.6–3.9 GW and Hanwha Group between 3.2 and 3.4 GW.

The firms’ manufacturing equipment which is dedicated to produce ingots, cells and modules or components and systems for solar PV farms is a stage in the PV supply chain which will not be addressed in this book. Unfortunately, the authors have not found any exhaustive economic/business study on this stage. Notwithstanding, the Website www.​enfsolar.​com contains a long and systematic list of those firms. In August 2015, this Website listed 1001 equipment manufacturing firms and 2103 components and system manufacturing firms. Almost all of them were located in the EU, USA, China, Japan, South Korea and Taiwan.

When they are off-grid plants, those professionals install the panels, the accumulator and the charge regulators.

These impacts are measured in detail and broadly by the photovoltaic life cycle analysis (see, for example, Fthenakis and Kim 2011; Gerbinet et al. 2014).

Simplifying somehow, funds refer to the production factors which provide their services in production processes, including labour force and machines, whereas the flows are the inputs that enter the process (inflows) and are incorporated to the output flow (outflows). This classification stems from a general descriptive model of production process in which the time dimension is explicit (see Mir-Artigues and González-Calvet 2007: 6–10 and 191–203).

It should not be forgotten that RD&D expenditures are not only incurred by private firms. Solar PV has also benefited from funds from public institutions.

The instant electromagnetic radiation from the external atmosphere is 1361 W/m² that, once impacting the earth crust, reaches an average of 1050 W/m² of direct sunlight and 1120 W/m² if diffused light is added.

According to one study from the US Energy Information Administration (EIA), cited in www.​photon.​info (September 2015).

With respect to the overall investment per unit of capacity, O’Rourke et al. (2010: 52) estimate that its value will go down from ~$5.5/W in 2009 to about $3/W in 2015 and $2/W in 2025.

In O’Rourke et al. (2009: 42), the values for c-Si modules are c = $0.3/W_p and f = $0.4/W_p, under the assumption that the price of p-Si is $50/kg (as was the case in 2010) and also assuming that they are vertically integrated firms, i.e. without margins due to the sales of the wafers and cells. For this reason, the price of c + m ≈ $1/W_p can be considered a value which is more in line with most firms in the sector.

This figure is inspired in Green (2005: 3) and MIT (2015: 31). It should be mentioned that the radius in €/W is the result of dividing the cost (€/m²) of the horizontal axis by the values in the y-axis, which represent efficiency. The number is obtained by multiplying this division by the inverse of the W generated per m².

Solar radiation changes much depending on the latitude and climate. For example, in the case of the Iberian Peninsula, the average annual solar radiation is greater than 5000 Wh/m²/day in the South Mediterranean coast and does not reach 3000 Wh/m²/day in the coast of the Biscay Bay. A typical module of one m² of 125 W_p under an insolation of 4 kWh/m²/day may generate up to 500 Wh/day. For values around Europe, see Šúri et al. (2006).

Currently, panels are particularly expensive when they are a few units which are integrated in the buildings as architectural components.

This information source expressed the capacity and the costs in terms of watts of alternating current. The values are in peak watts of DC and have been multiplied by 0.84. This circumstance does not influence the relevance of this information.

The quantity of water being required is limited: between 0 and 18.9 l/MWh (Drury et al. 2012: 10–46).

The debt service should not be disregarded for accounting purposes. This value, which includes the sum of the payment of interests and the repayment of the principal which was used to finance the initial investment, usually has an amortization period which is shorter than the lifetime of the plants (i.e. about 10–15 years versus at least twice that period).

Measuring this cost is an immediate task. See the discussion on the levelized cost of electricity (LCOE) below.

Something similar occurs with the efforts incurred by individuals and commercial premises when adopting energy efficiency and savings measures: the moving costs that the shift implies discourage it. Perhaps if the project was coupled to the service of moving furniture and appliances, the decision would be encouraged.

Ironically, the greatest advantage would be for families in a situation of energy poverty (>10 % of the revenues dedicated to purchase energy) which are nevertheless those which lack the capacity to invest.

Between 2012 and 2014, a moratorium paralysed all new renewable electricity capacity in Spain. This led to the proliferation of large PV projects whose electricity would be sold at pool prices. However, those projects do not seem to have materialized.

The emissions associated with solar PV generation have not been considered. All these emissions would be indirect, i.e. from the maintenance operations and supervision of the plants (Held et al. 2014). Its magnitude per kWh is very low, probably lower than the indirect emissions associated with conventional plants (due to the importance of the transport of fuel).

A medium-term analysis (several years) should take into account the evolution of the installed capacity (\(\Lambda\)). It should also not be forgotten that this greater capacity would lead to the location of some plants in worse places, and thus, the average \(\pi^{*}\) of the sector would go down slightly.

In addition, this prevents accessing one possible source of resources in order to finance FITs and other support schemes for renewable electricity. Thus, the burden of the policy costs has traditionally fallen on electricity consumers.

It should have been taken into account that growing quantities of solar PV electricity are used in the manufacturing of cells and panels.

It is assumed that investors would install the most efficient panels at each moment.

However, this diagnosis would be modified by longer useful lifetimes for the equipment and, especially, by the emergence and diffusion of new types of cells (and modules) which break the physical limits in the efficiency of the current technologies.

This conclusion is also applicable to exporting countries in the short and medium terms. In this case, the combination of high levels of solar radiation and crude oil reserves suggests that most of the oil should be sold in the world market if the cost of the conventional kWh is above the solar kWh.

In theory, those bids reflect the variable generation costs of the different plants although, in practice, the flexibility restrictions and the start-up costs of important generation plants and the existence of market power blur this influence (Murray 2009: 25, 46–48).

Strictly speaking, windfall profits are above normal profits, whose value can be linked to the regular profitability of financial funds.

The legal form of those operators can be highly diverse, although the transmission stage is usually in the hands of a public firm, whereas the distribution and commercialization activities are carried out by private firms.

The standard frequency (in Europe) is 50 Hz, with a deviation of ±0.2 Hz, whereas the voltage has to be maintained at ±5 % of its nominal value (Murray 2009: 181–182).

This literature is summarized in Hirth (2013, 2014).

A report has recently been published, IRENA (2015: 43), which assesses the integration costs of variable renewables as being between $0.035 and $0.05/kWh for 40 % penetration. These modest values significantly contrast with the negative (or very low) values which were estimated for low levels of penetration.

Support (in the form of FITs) should be added to this value. If only these existed, then the considered indicator could not be applied. On the other hand, it should be noted that the market prices are used as a reference, ignoring the multiple circumstances which may distort these prices, including market power. If the electricity market was perfect, then the market value would equal the marginal economic value that the solar PV has for society.

An alternative measure of the capacity value is the ECP (equivalent conventional power), which refers to the volume of thermal generation which should be added to the system in order for its LOLE to remain unchanged in case one unit of the referred generator had to be substituted.

It should be pointed out that solar PV plants are usually available during all daylight hours. In contrast to thermal plants, they do not suffer inactivity periods due to repairment and maintenance.

Depending on the latitude, the number of hours of solar radiation changes depending on the season.

The change in the level of radiation between two time intervals, measured with respect to the maximum of solar radiation, may show values higher or lower to 0.5 at 1-min intervals. This fact has a great impact on residential installations, but its influence is lower for utility-scale plants.

It should be recalled that solar PV generation lacks thermal inertia and cannot be used to mitigate the variability in the very short term.

On the diversity of reasons for and the economic effects of curtailment, see Kingle Jacobsen and Schröder (2012).

The use of physical magnitudes in the denominator is explained by the assumption that the energy generated by a plant goes in tandem with the revenues being obtained.

Given that this is too simple an approach, using probability distributions of the variables in the calculation of the LCOE in order to carry out Monte Carlo simulations has been proposed (Darling et al. 2011).

Spanish data on Wednesday, 17 July 2013. Source: www.​ree.​es.

It should be indicated that, on the one hand, the hourly values of the demand and the PV generation are expressed setting their respective absolute peaks at 100. On the other hand, the horizontal axis refers to solar time, and thus, a couple of hours need to be added in order to obtain the official hour. Then, the total sum of the normalized demand has been obtained and the total amount of the solar PV contribution has been proxied. Once the percentage of solar PV generation in each hour with respect to the total generation has been calculated, a new hourly solar PV curve has been built. Next, the effect of growing degrees of solar PV penetration has been simulated, knowing that the reading of the results is an immediate one, with the guarantee that the rescaled curves maintain the relative form and position of the original curves. Thus, the space between the upper curve (the initial demand curve) and the successive lower curves is the share of the electricity needs which are covered by solar PV electricity generation. To end up, it should be added that the total demand on Wednesday, 17 July 2013, was 738,819.5 MWh, with a contribution of solar PV of 30,525 MWh (i.e. 4.13 %).

In this section, electricity prices refer to wholesale electricity prices and not to final electricity (retail) prices. The later can be influenced by many other factors and/or are insensitive to peak situations, depending on the regulatory framework.

The data on the Mediterranean winter are from Spain (on Wednesday, 16 January 2013). Figures 4.14, 4.15 and 4.16 have been drawn based on data from Red Eléctrica de España (www.​ree.​es). Since the hours are in solar terms, an hour should be added in order to derive the official time. The data for the summer time correspond to the data from Fig. 4.12. Since the authors did not find real data for those dates in the case of a temperate climate, the curves shown are merely indicative of events in Atlantic Europe.

Ramping refers to the rate of change of net load between consecutive hours (hourly ramping value). Ramping is a problem when it is above the level at which the contribution of electricity may increase/decrease (see below).

The original data correspond to Wednesday, 16 January 2013, when the electricity demand was 799,054.6 MWh, of which 6989.6 MWh was covered by solar PV electricity (0.874 %).

Of course, the analysis could be extended ad infinitum. Spring and autumn seasons could also have been taken into account, as well as weekends, holidays, different weather conditions, etc. In fact, the description carried out would have specific features depending on all those variables, as well as the initial mix of the specific electricity system being assessed.

The notation used in this section in order to identify the parameters is not related to the notation in the rest of the book.

The difference between this area and the initial area is given by \(\frac{2}{3}\left( {d^{*} - d^{\prime} } \right)\left( {h_{b} - h_{a} } \right)\).

Gas and diesel-fired (peaker) plants have rapid ramping capabilities and low start-up costs (Borenstein 2012: 74). The instant reaction, however, only occurs with the hydro plants, in case they have enough water availability.

CESUR stands for “contracts for the last-resort provision of energy”.

The details of the operation of the day-ahead hourly market, as well as the operation of all the markets and associated activities mentioned in the text, can be consulted in Pérez-Arriaga (2013: 364–385).

The opportunity costs for a thermal plant are the lost profits for being unable to participate in the spot market (Hirth and Ziegenhagen 2013: 18).

Primary reserves represent 3 GW in Continental Europe, which are deemed enough to cover the loss of two nuclear reactors.

In the case of Germany, for example, plants receiving FITs cannot provide services in the control power markets although, since 2012, they can do it if they are remunerated at market prices plus a premium. However, although 40 % of the variable capacity can be selected for the control power markets, it never participates. The reason is that the regulation requires commitments for at least a week, a period over which the wind and solar predictions are not precise. Furthermore, the solar PV generation hours are below the amount of hours which have to be daily covered. Setting shorter dispatch intervals has been suggested (Hirth and Ziegenhagen 2013: 20, 2014: 15).

These losses are proportional to the square of the flux they involve.

It should not be forgotten that the value of the investment recovered by the project developer (as recorded in the accounting books) does not have to match the value allocated from the perspective of the electricity system.

In IRENA (2015: 42–43), it is estimated that for 30–40 % wind penetration levels, the impact on the transmission grids would be lower than $0.013/kWh. The same source reports that for low levels of penetration, wind and solar PV generation would lead to the avoidance of investments in distribution lines and, thus, to a small saving: between $0.003 and $0.007/kWh. For levels of penetration above 15 %, grids should be extended/reinforced, although the cost does not exceed $0.012/kWh.

Energy storage could play an arbitrage role in wholesale electricity markets: low-cost off-peak electricity would be purchased and be sold during on-peak periods. However, available bulk energy storage technologies cost more than $1000/kW. This cost does not justify the storage deployment for market arbitrage.