by Vaclav Smil
A further step toward landless solar electricity generation is the installation of PV walls. Thin-film PV cells made of copper indium gallium selenide can be laminated directly into walls (and, obviously, into roofing materials), and as their efficiency rises to rival that of the silicon-based cells, it will become more appealing to embed them into the south-facing walls of new buildings. Tall buildings offer the greatest opportunities but also experience significant reductions in insolation owing to shading, and wall PV installations would have suboptimal angles of irradiation. These realities are illustrated by a study of a pioneering PV wall installation, an integrated curtain wall facade that spans 12 floors of the lower part of the Solaire Building in New York, built in 2004 (Perez et al. 2012).
The PV array of monocrystalline silicon cells covers 153.5 m2, has a peak capacity of 11.3 kW, and faces the Hudson River waterfront, and hence its azimuth is 275 degrees. That is hardly an optimal orientation and one whose only advantage is a largely unobstructed exposure except for shading by a few trees. The building's wall receives 766 kWh/m2. The unshaded rate would be 822 kWh/m2, a horizontal area in the same location would get 1,430 kWh/m2, and the tilted south-facing surface would receive 1,615 kWh/m2, or 2.1 times as much as the actual wall installation, whose annual electricity generation was only 5,560 kWh/year (635 W), resulting in a power density of just 4.1 We/m2.
Concentrating Solar Power
Concentrating solar power (CSP) stations use tracking (computer-controlled) parabolic mirrors (heliostats) to reflect and concentrate radiation on a central receiver placed on a high tower. The concentrated radiation is then used to heat a transfer fluid (molten salt, whose temperatures reach up to 650°C), which then heats steam to power a turbogenerator. This technique has three obvious advantages when compared to PV plants: it can achieve higher conversion efficiencies; it can be used in a dual arrangement with fossil fuel or wood to generate steam during the night or during periods of higher demand; and a part of the peak heat flux can be stored in order to generate electricity at night or during periods of low irradiation, with molten salt as the best storage medium (Azcarraga 2013). Despite these advantages, the typical power densities of CPS-based electricity generation are not superior to those of PV-based electricity generation.
Solar One, the pioneering solar tower project designed by the US Department of Energy and located east of Barstow, California, generated electricity between 1982 and 1986 (CSP World 2012). Its field of 1,818 tracking heliostats covered 72,650 m2. The project was reopened in 1995 as Solar Two, with added heliostats and with the use of molten salt heat storage to smooth the fluctuating irradiation. Solar Two generated 17.5 GWh/year (an average rate of 2 MW) from the total area of 82,750 m2 of heliostats (USDOE 1998). The average power density of Solar Two was about 24 We/m2, but after only four years the plant was shut down, and in 2009 the tower was demolished and all heliostats were removed.
Europe's first commercial solar tower project, Spain's PS (Planta Solar) 10, completed by Abengoa Solar in Sanlucar la Mayor in 2007, is rated at 11 MWP and generates 24.3 GWh/yr, that is, 87.5 TJ/year at a rate of 2.77 MW (Abengoa Solar 2013; fig. 3.3). At 25%, its capacity factor is fairly high. Its heliostats occupy 74,880 m2 (624 x 120 m2), and the entire site is about 65 ha. This translates to a power density of about 37 We/m2 (for heliostats), and to a bit more than 4 We/m2 for the plant's total area; the latter rate is very similar to the performance of PV-based plants. PS20 (in operation since 2009) is rated at 20 MWP. It generates 48.6 GWh/year (175 TJ/year at a mean rate of 5.55 MW) and has a slightly higher capacity factor of nearly 28%. With mirrors covering 150,600 m2, the project's heliostat power density is 36.85 We/m2, almost identical to that of PS10, but at 6 We/m2 the rate for the entire site (about 90 ha) is nearly 50% higher.
The world's largest CSP project is Ivanpah Solar Electric Generating System (SEGS) in the Mojave Desert in San Bernardino County, California, a site with an exceptionally high annual irradiation of 2,717 kWh/m2 (310 W/m2). The project is owned by NRG Energy, Google, and BrightSource Energy, and the three fields have a total of 173,500 heliostats covering 260 ha (the entire project area is 1,400 ha) serving three 138-m-tall towers. The project's total gross installed capacity is 392 MWp and the expected annual generation is 1.079 TWh, that is, an average rate of 123.2 MW (BrightSource 2013). These specifications prorate to power densities of 47.4 We/m2 for the heliostat area and 8.8 We/m2 for the entire project area. The first rate is far higher than the power densities of the best currently operating PV facilities. No stunning improvements are foreseen for CSP efficiencies. This makes it safe to conclude that optimally located solar concentrating plants will generate electricity with power densities of 40-50 We/m2 of their large heliostat fields and with rates no higher than 10 We/m2 of their entire site area.
Figure 3.3
Abengoa central solar power plant. PA Pundits-International.
There is yet another choice to concentrate solar power, not by focusing sunlight onto a single point but by deploying large numbers of Fresnel lenses to concentrate sunlight (raising its intensity by two to three orders of magnitude) onto individual mulitjunction PV cells. The best efficiencies of these expensive cells are now in excess of 40%. Fthenakis and Kim (2013) prepared a life-cycle assessment of such a system, the Amonix 7700 highconcentration PV array in Phoenix, Arizona. This massive tracking unit has an area of 267 m2, and its installed peak capacity of 53 kW was expected to rise to 62 kW with improvements in the optical bath and better lens tuning. These specifications translate to a power density of 231.2 We/m2, an order of magnitude higher than that for nonconcentrating PV installations.
Potential Gains and Limits
Increased power densities will come with further gradual improvements in PV conversion efficiencies. News of such gains come regularly: in 2014 the record was held by Sharp's concentrator triple-junction compound solar cell, which used Fresnel lenses to concentrate radiation onto a layered cell made from, from the top, InGaP, GaAs, and InGaAs on a silicon substrate, at 44.4% (Sharp 2013). And the record for a commercially available multijunction cell with modular design (actually four cells stacked one on top of another) designed by Semprius (North Carolina), reached 35.5% (Semprius 2013), and the company believes that the efficiency could eventually rise to 50%. Another innovation that could boost photovoltaic's conversion rates involves cell coatings made from organic and inorganic semiconductors that maximize the harvesting of radiation (Tabachnyk et al. 2014).
But in the near terma advances in large-scale electricity generation will come from cheaper modules, not just silicon-based but also increasingly made from inexpensive thin-film perovskites (calcium titanium oxide, CaTiO3). In light of the history of continuing efficiency gains in PV-based electricity generation it is only a matter of time until the best annual power densities of PV conversions in large stations commonly surpass 10 We/m2, and it is not unrealistic to think that in 20-30 years, solar plants in the sunniest locations will routinely approach, and surpass, not just 20 but even 30 We/m2.
Significant power density gains would be realized with three-dimensional solar energy generation. Bernardi and co-workers (2012) explored such possibilities by modeling and building experimental 3D PV structures (3DPV) that combine absorbers and reflectors in the absence of sun tracking. Their three choices-an open cube, an open parallelepiped twice as tall as the cube, and a tower using slanted panels-could generate power densities that were 2-20 times higher per base area than those of stationary flat PV panels, that is, maximum rates in excess of 100 We/m2. In comparison, the gain for a flat panel with dual-axis tracking would be only 30%-80%. Of course, this increased density required a larger cell area (by a factor of 1.5 to four compared to flat panels), but this drawback is more than compensated for by other advantages of the 3D designs: compared to flat stationary panels they can double the hours of peak power generation, and they can greatly reduce seasonal, latitudinal, and weather variations. When combined with inexpensive thin PV films they
could open up new possibilities for large-scale PV-based electricity generation.
Another advantage of PV systems is their relatively high safety. Fthenakis and Kim (2011) studied material and energy flows in four commercial PV designs, those of monocrystalline silicon, multicrystalline silicon, ribbon silicon, and cadmium telluride. They concluded that the PV cycle is much safer than conventional energy sources in terms of both statistically expected and possible maximum consequences. At the same time, a German-like mass installation of both rooftop PV units and large multimegawatt projects is not applicable to all kinds of environments and to all levels of economic development. Germany is not an obvious choice for the world's largest installed PV capacity (the average irradiance over large parts of southern Spain and Italy is nearly twice as high), but the country's combination of advantages will not be replicated anywhere else anytime soon.
Germany has a modern electrical grid and distribution system, diverse manufacturing capabilities, and technical prowess, all of which made it fairly easy to ramp up the production of modules (which was later undercut to a large extent by cheap, subsidized imports from China) and facilitated the production and installation of the needed infrastructure for distributed PV-generated electricity (panel frames, wiring, inverters, meters, and increasingly also storage batteries). Germany's rainy climate provides natural cleansing of modules, and the traditionally fairly high electricity prices have made even early PV-based technologies relatively more competitive than in countries with inexpensive electricity. Also, a large segment of the German population prefers to adopt, and is willing to pay for, for more expensive renewable energy sources. And, of course, generous feed-in tariffs with prices guaranteed for two decades have offered a sure way to make PV-based electricity generation profitable for anybody who has the initial capital investment.
This combination explains why the pace of rooftop PV advances has been much slower in countries where even a few of Germany's advantages are missing: American sensibilities are not as green, and the Americans have always paid much less for their electricity, but a large part of the United States has irradiance twice as high as Germany's, and its high-voltage grid, electricity distribution, manufacturing potential, and technical skills are not inferior. But thanks to high feed-in tariffs, in 2012 Germany had in per capita terms 16 times as much PV capacity as did the United States (Fraunhofer ISE 2012). And replicating Germany's PV achievements would be outright impossible in countries with dodgy grids and unreliable electricity distribution (including a large share of illegal hookups).
It would also be a challenge in nations with a shortage of the technical skills needed to deploy the PV infrastructure for millions of units; in climates where seasonally heavy deposition coats the modules in dust and where wind-driven sand pits module surfaces; in societies where subsidized energy prices have created unrealistic expectations about the cost of the future energy supply and where the abundance of domestic energy resources does not create such urgency as does Germany's high dependence on fossil fuel imports; and in economies where only very few people could afford the initial investment in rooftop PV units, even it was assumed that house roofs were accessible and free for installation: in many densely cities of the Middle East and Asia they are not because they are covered either by illegal structures or by rubbish.
Further, systems considerations dictate that major shares of PV-based electricity generation (between 5% and 15% of the total supply; in 2013 Germany derived 4.7% of its electricity from PV-based generation) will be possible only with greatly expanded storage, a strategy that is now pursued both by Germany and in California. In Germany the program, started in May 2013, is limited to a small PV system of up to 30 kW, and a subsidy is offered for up to 30% of the price of storage systems tied to new or existing PV units. In California the three major utilities will have to buy 1.325 GW of storage capacity by the year 2020.
Wind and Wind-Generated Electricity
Only a tiny fraction of solar energy reaching the Earth goes into energizing the global atmospheric circulation, and hence the aggregate power of exploitable wind is orders of magnitude smaller than that of the planetary irradiance. Still, some recent estimates have concluded that it is many times larger than the global total primary energy supply (TPES). Archer and Jacobson (2005) assessed the global wind power potential at 72 TW (compared to the 2012 TPES of nearly 17 TW) even if only 13% of the Earth's windiest regions were exploited. Lu, McElroy, and Kiviluoma (2009), assuming a larger area and more powerful turbines, put the total land-based potential nearly 75% higher, at 125 TW. Later, Jacobson and Archer (2010) implied that more than 170 PW of wind power are available for extraction in the atmospheric boundary layer region.
Such high totals are in line with the arguments by Roberts and co-workers (2007), who claimed that the power that could be extracted from the planetary jet streams is one to two orders of magnitude greater than that extractable by equal-sized ground-based wind turbines, and proposed that a tethered rotocraft be used to extract this enormous flux, reaching horizontal power densities up to 20 kW/m2, compared to less than 500 W/m2 for a large modern ground-based turbine. Miller, Gans, and Kleidon (2011) exposed the fallacy of these high estimates by pointing out that all of them neglected energy conservation, and proceeded to derive a realistic estimate first by tracing the fundamental top-down process of energy transfer.
About 25% of the incoming solar radiation (45 PW) creates pressure differences due to differential heating (Lorenz 1976). About 2% of that total (900 TW) is the maximum available for wind power extraction; half of that total is dissipated in the atmospheric boundary layer (Peixoto and Oort 1992), and 25% of the remainder (112 TW) is dissipated over land, of which no more than 60% (68 TW) could be extracted over nonglaciated terrain. Miller, Gans, and Kleidon (2011) then proceeded to refine this estimate by adding a simple momentum model with reanalysis of wind data and by using climate model simulations. They concluded that the maximum wind power that could be extracted from the atmospheric boundary layer over all nonglaciated land (limited by the rate of its generation in the climate system) is, depending on the estimation approach, as low as 18 TW, and no higher than 68 TW. The lower estimate would prorate to 0.15 W/m2 of icefree land, the higher one to 0.5 7 W/m2, both being merely theoretical maxima of extractable kinetic energy.
Similarly, Adams and Keith (2013) used a mesoscale model to demonstrate that wind power production is limited to no more than about 1 W/m2 for any large-scale wind farm with turbines spaced over an area larger than 100 km2. Obviously, in practice only a very small share of the Earth's wind energy that could be captured will eventually be converted to electricity by commercially viable turbines. I will assert, with a high degree of confidence, that tethered rotocrafts will not be delivering electricity on a terawatt-hour scale anytime soon, and that the vast continental regions with very low wind speed and with seasonal doldrums will not see any installations of extensive wind farms. Hoogwijk, de Vries, and Turkenburg (2004) used annual wind speed data from the UK's Climate Research Unit to put the economic potential (cutoff at about $1/kWh) at 96 PWh/year (10.96 TW).
The best assessed new total of global wind capacity comes from a study that combined a reanalysis of wind speed data with assumptions about updated wind turbine performance, land suitability factors, and average costs, including those of requisite long-distance transmission (Zhou et al. 2012). Its central assumptions resulted in a total global economic (at less than 9 cents/kWh) wind generation potential of approximately 119.5 PWh/ year, or 13.6 TW. Sensitivity analyses show that the estimates depend particularly on assumed wind speed (ranging from - 70% to +450%), land suitability (from - 55% to +25%), turbine density (from - 60% to +80%), and cost and financing options (from - 20% to +200%).
The central estimate-13.6 TW, or just 0.11 W/m2 of ice-free land-is obviously depressed as a result of large continental areas with low wind speeds, particularly common during extended high-pressure spells in the northern hemisphere. In contrast, t
he power densities for windy coastal and inland regions will be at least an order of magnitude higher. Wind's considerable vertical power densities (Kapitsa's Umov-Poynting vector, explained in the second chapter of this book) translate into relatively low (horizontal) power densities of electricity generation even at the windiest sites. I will illustrate this progression of values by using actual data for one of the larger machines on the market, the Vestas model V90, which has a rated power of 3 MW (Vestas 2013).
The turbine begins generating once wind speed reaches 3.5 m/s, its cutout is at 25 m/s, and its rated wind speed (v) is 15 m/s. With the rotor diameter of 90 m, the three blades sweep (with the nominal revolution of 16.1 rpm) an area (A) of 6,362 m2. These ratings (and air density p = 1.2 kg/m3) set the maximum power (P) of this Vestas turbine at 12.88 MW:
The Betz limit (0.59; Betz 1926) takes that down to a maximum achievable rating of 7.6 MW, and the actual rated power of 3 MW means that the machine has a fairly high correction factor (0.39, owing to performance losses in mechanical energy conversions) and that the Umov-Poynting vector (vertical energy flux) of its electricity generation is 471.5 W/m2 of the area swept by its blades.
Box 3.3
Maximum power of Vestas wind turbine
Power Densities
Compared to this fairly high rate, the power densities of wind-driven electricity generation as energy flux prorated per unit of horizontal surface are two orders of magnitude lower because wind turbines must be set apart to avoid excessive wake interference. Turbines must be placed at least three, and better five, turbine diameters apart in the cross wind direction, and at least six and preferably ten diameters in the downwind direction (Hau 2005). The power density of large wind farms with turbines on a square grid is thus 2-3 W;/m2. Returning to the Vestas 90, spacing these machines in a large wind farm on a regular rectangular grid five and ten rotor diameters apart would result in a power density of 2.37 We/m2 for the rated performance of 3 MW;/turbine. The American Wind Energy Organization lists the power densities of more than 70 large-scale (what they call industrial) facilities on its website, with the rates ranging from just below 1 W;/m2 to 11.1 Wi/m2 at an exceptionally windy site (Cassia County, Idaho) and with most densities, as expected, between 2.5 and 4 Wi/m2 (AWEO 2013).