Page:Quantifying a realistic, worldwide wind and solar electricity supply.pdf/10

 Y.Y. Deng et al. / Global Environmental Change 31 (2015) 239–252

248

Table 9 Potential for electricity generation by technology and period. Potential (EJ/a)

Period

Buildings

On-shore

Off-shore

Country meta-study

PV on roofs

PV on facades

PV

CSP

Wind on land

Wind on sea

Geoth. elec.

Hydro

90–804 166–1,480 316–2,815

98–808 118–970 131–1,078

25–170 27–185 27–185

189–624 197–652 197–652

1 3–8 27–84

12 16 26

n/a

n/a

n/a

n/a

118–1109

53

Single technology

2010 2030 2070

31 64 121

7 15 90

Total potential on land (all technologies), accounting for overlaps

2010 2030 2070

n/a

n/a

Potential ranges from literature

Technical (constrained) potential

n/a

n/a

132–1,112 172–1,497 320–2,832 1340–14,780

(Edenhofer et al., 2011). Where our results are smaller than the SRREN maximum potentials ranges (i.e. for PV and CSP), this is primarily due to our stricter constraints on land availability. The only exception is off-shore wind where we ﬁnd a much larger potential than SRREN. 85% of our potential is found at depths over 200 m. This goes beyond the typical depths used today ($40 m) and the majority of the studies surveyed in the IPCC’s report. 4.2. Additional on-shore wind potential locally There is uncertainty over where to place the threshold for excluding areas based on available wind resource. This is due to the wide range of local wind speeds local orography can result in but also the uncertainty in some geographies of our wind speed data set (which nevertheless has a higher resolution than that used in several previous global studies). Because our motivation was to ﬁnd reliable potential estimates, we have used a value for the threshold which is at the lower end of current practice ($6 m/s at hub height), meaning we may have excluded areas which may have local potential. In addition, in the time horizon used in this study, more areas with even lower average annual speeds are likely to become attractive for developers, as various constraints force projects into less resource rich locations.

Fig. 3. Global potential results. Annual achievable electricity potential is shown by technology for 2010 and 2070. Shading within each technology shows the Low, Medium and High availability cases. The lines show indicative estimates of future demand for a global population of 10 billion (IEA, 2012a; GEA, 2012; Lund and Mathiesen, 2009; Deng et al., 2012; Shell, 2013).

250–10,790

350–1800 (70–220)

All wind: 1000–3050 (140–450)

It should also be noted here that our assumption on the power density for wind of 7 MW/km2, while in line with current practice and previous studies, has been called into question recently, for wind farm installations spanning hundreds of square kilometres (Adams and Keith, 2013) due to turbine interference beyond the level we have assumed here. In such very large single farms, achievable power densities as low as 1 MW/km2 have been suggested. Our availability assumptions on land imply a small likelihood of a necessity for very large single wind farms. For offshore wind, our results are slightly larger than comparable studies (EEA, 2009; Schwartz et al., 2010). This is primarily due to the limits other studies place on ocean ﬂoor depth, usually around 50–200 m maximum: the current predominant off-shore wind technology requires anchoring on the ocean ﬂoor. However, innovative technologies, such as ﬂoating turbines or similar, have the potential to extend our reach and we have included areas as deep as 1000 m to reﬂect this potential. This highest depth class represents around 160– 550 EJ/a, equivalent to $85% of the total offshore wind potential in all cases. 4.3. Buildings could host over a third of PV potential For building-based PV, the approach in this study is broadly consistent with other studies. Based on ﬂoor area, an estimate of sun-facing roof and fac¸ade area is obtained, followed by a suitability factor. For the resulting available area the annual irradiation is multiplied by the system efﬁciency of the solar panel which results in the net potential. We ﬁnd a potential of 210 EJ/a in 2070, compared with 38 EJ/a in the base year (2010). There are few global studies with similarly detailed approaches to compare to. Compared to other national studies the 2010 potentials we ﬁnd are generally larger. For example, a study for PV on roofs in the United States reported 2.9 EJ/a (Denholm and Margolis, 2008) while this study reports 4.6 EJ/a and a study for China found 1.9 EJ/a (Zhou et al., 2009) while we report 5.2 EJ/a. Assumptions are often not clearly stated, but the differences are likely due to the exclusion of rural roof areas or different available area and suitability factor estimates. When comparing our potential estimates for PV on buildings with the low end of the range of our estimates for PV potential on land, we ﬁnd that buildings could contribute as much as a third of PV potential and could thus make a meaningful contribution to solar renewable electricity provision. This could warrant focussing support policies in this area, especially considering that buildingbased PV does not require additional land, is already costcompetitive with conventional sources in some regions and has the potential to alleviate system costs by reducing peak electricity demand.