City-integrated renewable energy for urban sustainability

While we are happy to engage in scientific debate, Miller et al. must recognize that the Kammen and Sunter article primarily is a review article. If the scientific concerns are with the content of the cited peer-reviewed publications, we suggest that Miller et al. pursue discussions with the authors and editors of those publications. Through these discussions if any of the cited publications are retracted, then please do let us know so the review can be updated accordingly.

The upper limit presented for the power density of solar photovoltaics is clearly stated in the Kammen and Sunter article as being the "potential performance, based on optimal conditions and technologies currently available in the laboratory." The Solar I/II power plant was included in the last rebuttal to show how even a slightly newer construction (started in 2013) had significant increases in power density, 26% increase when compared to the average of the three plants cited by Miller et al. Photovoltaic efficiencies have improved greatly. Industrial-scale solar farms constructed over 5 years ago are clearly not using "technologies currently available in the laboratory". The Kammen and Sunter article is forward looking at where the potential for the photovoltaic industry could be in upcoming years as these laboratory technologies are brought into the market, not backward-looking at where the industry was 5 years ago.

Miller et al. state that "the math here is obvious, but incorporating such estimates into practical policy decisions is irrelevant." We strongly disagree. Forward-thinking policies that prepare for a dynamic and changing world are not only relevant and practical but critically important. Policy decisions regarding where to invest research and development funding, how to structure building codes and permits that ensure safe installations of renewable energy technologies, and city planning efforts that design city infrastructure that minimize energy consumption and optimally use local energy resources are just some of the many areas that could benefit from understanding the current status of cutting-edge research in renewable energy technologies. Even if Miller et al. find the work irrelevant, clearly there are others that do not as this article was downloaded over 2,000 times within the first two weeks of it being published.

We are uncomfortable with Miller et al. suggesting agreement on statements we have never made. Miller et al. wrote "we agree that if all surfaces (roads, buildings, parks) in the city were completely blanketed with solar PV that operate at a 40% efficiency, one could achieve 120 We/m^2 in many locations." Nowhere in our paper or our rebuttal have we suggested that all surfaces be completely blanketed with photovoltaics. Certainly, solar installations do not need to be limited only to rooftops. In the Kammen and Sunter article, several studies are cited that have explored the opportunities associated with other installation locations, particularly building façades; yet, "the risk of vertical obstruction and shading" is clearly mentioned. A potential power density of up to 120 We/m2 does not imply that all square meters of the city be covered with photovoltaics. We introduce several cutting-edge technologies. A combination of these should be chosen based on the available natural resources of a given city and installed where most appropriate.

With regard to the wind energy estimates, Miller et al. attempt to make their point by appealing to a rudimentary caricature of the fluid mechanics that lacks essential features of the governing physics. As but one example, Miller et al. estimate the wind energy resource by using a referenced value for the "friction velocity of a city". In fact, there is no single "friction velocity of a city". The concept is ambiguous at best and meaningless at worst, as the friction velocity in a given flow depends on details of the pressure and viscous drag on objects in the flow. These forces are specific to the topography and flow conditions of a given location (e.g. the wind speed, wind direction, thermal stratification, building height, building shape, building spacing etc.). Indeed, the authors of the paper cited by Miller et al. note themselves that "the land-surface scheme in the model needs improvement to resolve the urban canopy layer and the roughness sublayer" (page 6552). Since Miller et al. do not resolve the urban canopy layer and the roughness sublayer, their estimates are not useful.

Rather than work to achieve the needed improvements to the model, Miller et al. fall back to a simplistic calculation of the energy fluxes and incorrectly use that straw man to show a limited wind energy resource in cities. Unbeknownst to them, Miller et al. actually demonstrated the veracity of the estimates in the Kammen and Sunter article via the model in figure 1b of their recent letter. That model is adequate for perhaps the first 100 meters of downwind distance, during which time the frictional effects of the city topography will be limited. In that regime, they find estimates for the wind energy resource that are consistent with those of Kammen and Sunter. Beyond this distance, their model is incorrect as it assumes a constant topography, a feature not found in any real city. A correct estimate of the downwind energy resource must balance the mean and turbulent fluxes of kinetic energy from upwind and above the city; the energy extraction of the wind turbines; and the dissipation of kinetic energy by turbulence created by the various classes of structures in the built environment, including the buildings, trees, and the wind turbines themselves. Once the authors forego their conceptual shortcuts and take the time to compute these parameters rigorously for a real city, they will find that their models are in agreement with the measurement-based estimates presented in the Kammen and Sunter article. Until then, further rebuttals by Miller et al. are not being made in good faith.

In conclusion, we are glad that through these eLetters Miller et al. have come to agree that "there is significant room for further renewables integration in cities" since demonstrating this potential was the intent of the article.

We challenged Kammen & Sunter [1] because their city-scale estimates were too high. Industrial-scale solar farms currently generate less than 20 We/m^2 (Kammen & Sunter noted 10-120 We/m^2) and city-scale wind farms could generate no more than about 1 We/m^2 (Kammen & Sunter noted 2.5-30 We/m^2). When compared to the ~28 W/m^2 average consumption rate of the 10 most populous cities, the 29 W/m^2 of Berkeley (CA), or the 47 W/m^2 of Cambridge (MA), hopefully our reason for forcing this debate is immediately clear.

We are trying to establish practical estimates for city-integrated solar PV and wind. Reflecting on our debate thus far, we are converging on solar. The opposite is true for wind, where their revised estimate of 35 We/m^2 is higher and therefore more wrong than before.

Figure 1a shows the average consumption rates of the 10 most populous cities in the US, as well as the chronology of solar PV and wind estimates raised in this debate so far.

Kammen & Sunter's original manuscript [1] stated a solar PV range of 10-120 We/m^2. We responded with observed solar farm data that was perhaps overly precise at 8.7-13.2 We/m^2, as Sunter, Dabiri, and Kammen then responded with another observation that was just above our range. Solar farms on flat ground could hit 20 We/m^2 in the very near-future, but the estimate is only part of the story here. We would be quite cautious in using any operational solar farm data as applicable to city-scales, as cities will exhibit complex shading considerations, more difficult operation and maintenance, and competing uses for land.

We agree that if all surfaces (roads, buildings, parks) in the city were completely blanketed with solar PV that operate at a 40% efficiency, one could achieve 120 We/m^2 in many locations. The math here is obvious, but incorporating such estimates into practical policy decisions is irrelevant. Reconfiguring a city to generate just the 20 We/m^2 noted above should be recognized as a monumental challenge, even leveraging the recent drops in PV manufacturing costs.

On wind, Kammen & Sunter [1] originally noted a range of 2.5-30 We/m^2. We noted a range of 0.3-1.5 We/m^2, to which Sunter, Dabiri, and Kammen responded with 35 We/m^2.

We reiterate that the limit to city-scale wind power generation has little to do with the specifics of the technology and rather relates to considerations of spatial scale. We recognize that [2,3] observed generation rates of 21-47 We/m^2, where in [2] the total generation rate is "Averaged over the 48.6-m^2 footprint of the six-turbine VAWT [vertical axis wind turbine] array…". For reference, New York City is about 784 000 000 m^2, and it is very much unlike an experimental site that is "vacant desert …[with] topography [that] is flat for approximately 1.5 km in all directions" [2]. Moreover, as stated in [3], "[the] unexpected, even provocative conclusion [of ~35 We/m^2 from 6-24 10-meter tall 1.2 diameter vertical axis wind turbines in various arrangements being applicable to larger scales] awaits confirmation either experimentally in larger-scale field studies or numerically in large-scale simulations that account for how the atmospheric boundary layer adjusts to the presence of the VAWT farm."

Such energetically consistent large-scale simulations, which include how the atmosphere will respond to wind turbines near the surface, formed the basis for our estimate of 0.3-1.5 We/m^2 [4-9]. These results from various researchers show that large-scale energetics prevents generation rates of anywhere near 35 We/m^2 from occurring at scales of 350-1600 km^2, reflecting the areas of the 10 most populous cities in the US.

There is no doubt that a single isolated turbine, or a line of turbines aligned perpendicular to the wind direction, could be influenced by a horizontal kinetic energy flux well over 70 W/m^2, if one considers only the physical footprint of the wind turbines. For cities, while the kinetic energy entering the upwind boundary of the city could be quite large, many cities extend downwind for 20 to 40 km (Philadelphia, Houston). At this downwind edge, the downward kinetic energy flux maintains the winds. Cities with wind power would utilize this entire expanse, so how can we estimate their limit to large-scale wind power generation?

As we know that energy must be conserved within the system, kinetic energy entering the city from the side and above provide a hard limit to how much electricity could be generated.

The horizontal KE flux into the city is: 0.5•ρ(v^3)•H•W, where the air density is ρ=1.2 kg/m^3, the wind speed v =5 m/s, the height is H ≈ 200m, and W is the width (m). The vertical flux downward into the city is ρ((u*)^2)•v•L•W, where the friction velocity of a city is about 0.5 m/s [10], and L is the downwind depth of the wind farm (m), referenced from the upwind boundary.

Figure 1b shows how the kinetic energy flux changes with the downwind depth (assuming each city can be approximated as a circular area), including downwind extents of the 10 most populous US cities. Note the 70 W/m^2 of kinetic energy entering the city at the upwind boundary – this could generate the 35 We/m^2 noted in the previous eLetter response by Sunter, Dabiri, and Kammen. By 1km downwind, the kinetic energy influx has dropped to 10 W/m^2, to 3 W/m^2 at 5km downwind, and to 1.5 W/m^2 by about 20 km. This 20km depth is approximately Philadelphia, the smallest of the 10 most populous US cities. This 1.5 W/m^2 is also in agreement with long-standing energetics-based estimates on what maintains our present-day climate [11,12].

Assuming half of the kinetic energy could result in electricity generation, this simple approximation suggests that no more than about 1 We/m^2 could be generated over the footprint of the city, regardless of the technology or their density.

Conclusion

We applaud the amazing solar PV cost reductions over the past few years. It is unequivocal good news and may be the most instrumental enabler to propel PV to being a major contributor to power generation. Assuming one could transform Phoenix, San Diego, or San Jose into commercial-scale solar farms, then these cities might be able to meet their average energy consumption rates. However, wind power could not meet the energy consumption rates of any of these cities. Other cities like Chicago, Philadelphia, and New York City seem unlikely to achieve renewable energy independence using either, or both, wind and solar.

There is significant room for further renewables integration—especially solar—in cities. But the greater opportunity for renewables like wind and solar resides outside the boundaries of the cities, as well as connecting cities to distant sunny or windy locations with high voltage transmission. This alternative approach, with sufficient large-scale deployment of solar PV or wind turbines, could power cities, and hopefully even yield a positive economic return on the investment into our collective future.

[FIGURE CAPTION]
Figure 1. a) In red, mean average energy consumption rate in 2014, estimated by deriving an average per capita energy consumption rate for each state [13,14], scaling this to the city population, and dividing by the city's area [14]. The blue text shows the chronology of the estimated wind (blue) and solar PV (black) generation rates. b) Assuming the same climatology for all locations and the incoming kinetic energy flux is full available (no wind power deployed upwind), showing how our simple kinetic energy budget estimates a kinetic energy flux into the upwind edge of a city at more than 70 W/m^2 for very small depths (10-100s of meters) but drops to 2-3 W/m^2 at downwind depths typical of major US cities. Our assumption is that about half of the kinetic energy influx shown on the y-axis could be extracted for electricity generation.

References

[1] Kammen D, Sunter D (2016) City-integrated renewable energy for urban sustainability. Science. 352, 6288, 922-928
[2] Dabiri J (2011) Potential order-of-magnitude enhancement of wind farm power density via counter-rotating vertical-axis wind turbine arrays. Journal of Renewable and Sustainable Energy, 3, 043104
[3] Dabiri J (2014) Emergent aerodynamics in wind farms. Physics Today. 67, 66-67
[4] Keith DW, et al. (2004) The influence of large-scale wind power on global climate. Proc. Natl. Acad. Sci. USA 101 (46): 16115-16120
[5] Miller LM, Gans F, Kleidon A (2011) Estimating maximum global land surface wind power extractability and associated climatic consequences. Earth Syst. Dynam. 2:1-12
[6] Jacobson MZ, Archer CL (2012) Saturation wind power potential and its implications for wind energy. Proc. Natl. Acad. Sci. USA
[7] Adams AS, Keith DW (2013) Are global wind power resource estimates overstated? Environ. Res. Lett. 8:015021
[8] Marvel K, Kravitz B, Caldeira K (2013) Geophysical limits to global wind power. Nature Climate Change. 3, 118-121
[9] Miller LM, et al. (2015) Two methods for estimating limits to large-scale wind power generation. Proc. Natl. Acad. Sci. USA 112 (36) 11169-11174
[10] Luhar A, Venkatram A, Lee S (2006) On relationships between urban and rural near-surface meteorology for diffusion applications. Atmospheric Environment. 40, 6541-6553
[11] Lorenz E (1955) Available potential energy and the maintenance of the general circulation. Tellus. 7. 271-281
[12] Peixoto J, Oort A (1992) Physics of climate. American Institute of Physics. Springer-Verlag, New York (USA)
[13] U.S. Energy Information Administration. Primary energy consumption estimates for 2014. http://www.eia.gov/state/seds/sep_sum/html/pdf/sum_btu_totcb.pdf (average state energy consumption rate in 2014, accessed July 14, 2016)
[14] US Census Bureau, QuickFacts United States. http://www.census.gov/quickfacts/table/PST045215/00 (city area and city population data current as of July 14, 2016)

Kammen, Daniel M., and Sunter, Deborah A. (2016) "City-integrated renewable energy for urban sustainability," Science, 352, 922 – 928. DOI 10.1126/science.aad9302

By Deborah Sunter, John Dabiri, and Daniel M. Kammen

The advances in renewable energy power densities have been immense. While such great achievements may be jarring, all the power densities brought into question have been peer-reviewed and are part of a new frontier for science.

In response to our paper Miller et al. states that the power density range for wind is "wrong" based on the global estimates of a limit to wind energy generation. However, these global estimates are based on the assumption of a fully developed atmospheric boundary layer over the wind farm, sometimes referred to as the wind turbine atmospheric boundary layer (WTABL). While even a first-principles analysis of the fully developed WTABL regime predicts that an order of magnitude improvement over current HAWT farm power densities is feasible (1), this flow regime does not most accurately represent wind behavior in urban areas. This condition of a fully developed WTABL is only achieved if the surface roughness is constant in the downwind direction over very long distances, i.e. if there are is a consistent topography over that length scale. In the letter to the editor, the authors even state a required distance downwind of "10km." For a wind farm in the plains, that condition could be achieved, e.g. if all of the turbines are similarly spaced and have a similar height over the full extent of a very large wind farm. However, a fully developed WTABL is prevented if there are changes in topography. In cities, the variable height and spacing of the built environment causes substantial changes in the topography. The fully developed WTABL can also be prevented if the turbines themselves are arranged such that the topography in the wind farm is variable, e.g. by using wind turbines of variable heights and/or spacing, as is demonstrated in a recently published lab-scale study (2).

When the WTABL is not fully developed, it is instead described as operating in a developing flow regime. In this developing flow regime, the energy that can be extracted is not limited in the manner referenced by the letter to the editor. Full-scale field tests of arrays of up to 24 vertical-axis wind turbines (VAWTs), an array size that could very well be implemented in a city, measured peak power densities over 35 W/m2 (3). Additional field tests of VAWTs with more conservative spacing measured consistent results (4). These directly measured field tests indicate the potential performance of wind turbines based on optimal conditions using laboratory tested technologies and, therefore, are shown accurately in Figure 1.

Next, the letter to the editor states that the power density range for solar photovoltaics is "misleading" and include the power densities of a few industrial solar power plants. We are concerned that the three selected power plants are not representative of the most up-to-date activities, even within the commercial space. The Agua Caliente Solar Project began construction in 2010 and portions of the power plant were online by January 2012 (5). The Topaz Solar Farm began construction in 2011 with the first solar panels being install in 2012 (6). We assume that the reference to the Central Valley Solar Ranch was meant to be the California Valley Solar Ranch that was first online in 2013 (7). Left out was the Solar I/II power plant with a higher power density of 14.6 W/m2 that was more recently completed in 2015 (8–10). In a quickly advancing field such as photovoltaics, technologies that were installed on power plants five years ago are neither "current" nor "state-of-the-art."

The letter to the editor states that a typical value of 10 W/m2 for photovoltaics is "not conservative." The value is based on a widely cited publication from 2009 that uses an average direct solar irradiation of 100 W/m2, which is typical for the United Kingdom, and a photovoltaic efficiency of 10% (11). The United Kingdom is not a particularly solar rich country with an average solar insolation only a third the global maximum (12). The global annual average horizontal surface solar irradiation is approximately 170 W/m2 (13). Additionally, the efficiency of an average commercial wafer-based silicon modules is 17% currently (14). The typical value of 10 W/m2 is calculated using below average values for both the solar insolation and commercial photovoltaic efficiency and, therefore, is conservative.

It is important to realize that all of the power densities presented in our Figure 1 (Sunter and Kammen) isolate each renewable energy source. There have been important additional efforts in hybrid renewable energy system (HRES) design which combine two or more renewable sources. PV/T systems combine photovoltaics with solar thermal. There are several commercially available PV/T systems and active on-going research (15). Another area of active research is PV/Wind hybrid systems (16). Rooftop PV/Wind integrated modules recently became commercially available (17). It is important to realize that cities are three dimensional and energy can be extracted at various heights increasing the total power density. For example, solar energy generation by photovoltaics on the rooftop certainly does not prevent geothermal energy from being extracted below the building. City-specific solutions are needed to optimally integrate multiple renewable energy technologies.

City-integrated renewable energy is not only a possibility; it is a reality that has been installed in cities around the world.

Dr. Deborah Sunter and Daniel M. Kammen, Professor
Renewable and Appropriate Energy Laboratory
Energy and Resources Group, and Goldman School of Public Policy
University of California, Berkeley
E: [email protected]; [email protected]
URL: http://rael.berkeley.edu

John O. Dabiri, Professor
Departments of Civil and Environmental Engineering and in Mechanical Engineering 
School of Engineering, Stanford University
E: [email protected]
URL http://dabirilab.com

1. P. Luzzatto-Fegiz, C. P. Caulfield, An ideal limit for te performance of a large, fully-developed wind farm. Bull. Am. Phys. Soc. Div. Fluid Dyn., BAPS.2014.DFD.E24.5 (2014).
2. A. E. Craig, J. O. Dabiri, J. R. Koseff, Flow kinematics in variable-height rotating cylinder arrays. J. Fluids Eng. (2016), doi:10.1115/1.4033676.
3. J. O. Dabiri, Potential order-of-magnitude enhancement of wind farm power density via counter-rotating vertical-axis wind turbine arrays. J. Renew. Sustain. Energy. 3 (2011), doi:10.1063/1.3608170.
4. J. O. Dabiri, Emergent aerodynamics in wind farms. Phys. Today. 67, 66–67 (2014).
5. Agua Caliente Solar Project, Arizona, United States of America, (available at http://www.power-technology.com/projects/agua-caliente-solar-project-ari...).
6. First Photovoltaic Solar Panel Installed on Largest Solar Project in the World (2012), (available at http://investor.firstsolar.com/releasedetail.cfm?ReleaseID=674965).
7. EIA, California Valley Solar Ranch, Annual Electricity Data (2015), (available at http://www.eia.gov/electricity/data/browser/#/plant/57439/).
8. EIA, Solar Star I, Annual Electricity Data (2015), (available at http://www.eia.gov/electricity/data/browser/#/plant/58388).
9. EIA, Solar Star II, Annual Electricity Data (2015), (available at http://www.eia.gov/electricity/data/browser/#/plant/58389).
10. Fact Sheet: Solar Star Projects. Sun Power (2014), (available at http://us.sunpower.com/sites/sunpower/files/media-library/fact-sheets/fs...).
11. D. J. MacKay, Sustainable Energy — without the hot air (2009).
12. NASA, Surface meteorology and Solar Energy Dataset 6.0 (2014).
13. World Energy Resources: 2013 survey. World Energy Counc. Chapter 8: (2013) (available at https://www.worldenergy.org/publications/2013/world-energy-resources-201...).
14. Fraunhofer Institute for Solar Energy Systems, "Photovoltaics Report" (2016), (available at https://www.ise.fraunhofer.de/de/downloads/pdf-files/aktuelles/photovolt...).
15. S. R. Reddy, M. A. Ebadian, L. Cheng-Xian, A review of PV–T systems: Thermal management and efficiency with single phase cooling. Int. J. Heat Mass Transf. 91, 861–871 (2015).
16. V. Khare, S. Nema, P. Baredar, Solar-wind hybrid renewable energy system: A review. Renew. Sustain. Energy Rev. 58, 23–33 (2016).
17. WindStream Technologies, SolarMill, (available at https://www.windstream-inc.com/my-solarmill).

Figure 1 of Kammen & Sunter places the energy demand of present-day cities (~20 W m^-2) between "typical [and]... potential [renewable energy] performance, based on optimal conditions and technologies currently available in the laboratory." Yet the Wind range of 2.5-30 W m^-2 is wrong, the Solar PV range of 10-120 W m^-2 is misleading, and cold weather downtowns often require up to 500 W m^-2.

After a wind farm extends downwind for more than ~10 km, including the consideration that wind turbines and the atmosphere will be directly interacting limits the electricity generation rate to about 0.3-0.5 W m^-2 globally, and about 1-1.5 W m^-2 in windy locations [1-6] atypical of cities. First recognized by [7], this limit relates to the atmosphere's ability to continually generate motion (~2 W m^-2 globally), rather than the specifics of the wind technology.

Solar PV, based on observed 2015 generation rates and footprints from 3 newly constructed solar farms, generates about 11 W m^-2 (8.7 W m^-2 over 2400 acres by Agua Caliente Solar Project; 10.5 W m^-2 over 3500 acres at Topaz Solar Farm; 13.2 W m^-2 over 1475 acres at Central Valley Solar Ranch, [8-13] consistent with [14,15]). 10 W m^-2 is, therefore, not "conservative" (p.922), but rather representative of state-of the art industrial facilities presently operating in regions of flat terrain with excellent insolation, and receiving ongoing maintenance. A doubling of PV generation to ~20 W m^-2 could be achieved if the entire city surface was PV – this does not violate physical laws, but is oblivious to economic trade-offs of cost vs efficiency. These physical and economic considerations explain why current PV generation rates in cities are typically lower than industrial sites.

Integrating renewables into cities is possible, and it might reconnect us to our environment with everyday reminders of our energy demands. A sound path begins by establishing achievable, near-term realities for renewables, and then incrementally leveraging these small successes.

Figure captions:

(a) Showing the result of estimating electricity generation rates from wind power which fail to consider (x-axis) or include (y-axis) how wind farms can slowdown wind speeds and alter atmospheric transport. Note how, for generation rates less than about 0.25 We m^-2, estimating generation rates with or without turbine-atmosphere interactions yield similar generation rates – these instances represent wind farms with larger between-turbine spacing, and therefore reduced interactions. The blue circles represent studies which did not include interactions, while squares represent studies which estimated generation rates with and without interactions. The red squares highlight those 'with turbine-atmosphere interaction' studies which estimated generation rates for wind farms in only windy locations [1,4,6]. All data points are the same as shown in Fig. 5 of [6].

(b) Observed monthly mean generation rate during 2015 for 3 industrial-scale solar PV farms in California and Arizona (USA) [8-13].

References

[1] Keith DW, et al. (2004) The influence of large-scale wind power on global climate. Proc. Natl. Acad. Sci. USA 101 (46): 16115-16120

[2] Miller LM, Gans F, Kleidon A (2011) Estimating maximum global land surface wind power extractability and associated climatic consequences. Earth Syst. Dynam. 2:1-12

[3] Jacobson MZ, Archer CL (2012) Saturation wind power potential and its implications for wind energy. Proc. Natl. Acad. Sci. USA

[4] Adams AS, Keith DW (2013) Are global wind power resource estimates overstated? Environ. Res. Lett. 8:015021

[5] Marvel K, Kravitz B, Caldeira K (2013) Geophysical limits to global wind power. Nature Climate Change. 3, 118-121

[6] Miller LM, et al. (2015) Two methods for estimating limits to large-scale wind power generation. Proc. Natl. Acad. Sci. USA 112 (36) 11169-11174

[7] Gustavson M (1979) Limits to wind power utilization. Science 204 (4388) 13-17

[8] Department of Energy (2010) Final environmental assessment for Department of Energy Loan Guarantee for Agua Caliente Solar Project in Yuma County, Arizona. DOE/EA-1797

[9] Harvey & Associates (2010) Biological assessment for the California Valley Solar Ranch Project San Luis Obispo County, California.

[10] Department of Energy (2011) Final environmental impact statement Volume I, Department of Energy loan guarantee to Royal Bank of Scotland for construction and startup of the Topaz Solar Farm San Luis Obispo County, California DOE/EIS-0458

[11] U.S. Energy Information Administration. Electricity data browser for Agua Caliente Solar Project, monthly. http://www.eia.gov/electricity/data/browser/#/plant/57373 (data for 2015 accessed on May 24)

[12] U.S. Energy Information Administration. Electricity data browser for California Valley Solar Ranch Project, monthly. http://www.eia.gov/electricity/data/browser/#/plant/57439 (data for 2015 accessed on May 24)

[13] U.S. Energy Information Administration. Electricity data browser for California Valley Solar Ranch Project, monthly. http://www.eia.gov/electricity/data/browser/#/plant/57695 (data for 2015 accessed on May 24)

[14] Smil V (2015) Power density: a key understanding energy sources and uses. MIT Press.

[15] MacKay DJ (2009) Sustainable energy -- without the hot air. UIT Cambridge.