2. General Constraints on Hydro, Geothermal, and Biomass Energy
The leading form of RE today is biomass, mainly fuel wood consumed in low-income countries. In 2017, it accounted for an estimated 9.5% of global primary energy 
. Although some researchers see modern bioenergy as a major future global energy source 
, strong competition from the other main uses for biomass—food, forage, fibre, and forestry—will heavily restrict its growth, as will the need to preserve biodiversity 
. Further, by 2050, the UN median estimated that global population would rise from the 2020 level of 7.8 billion to around 9.7 billion 
, and non-energy biomass needs can be expected to rise proportionately. Liquid biofuels in 2019 accounted for only 4.1 exajoule (EJ = 1018
joule) globally, with US and Brazil together supplying 62% of the total 
. Further, expansion of tree-cover for bioenergy plantations in northern climates, while drawing down CO2
, would tend to decrease local albedo (by replacing high-albedo snow cover with lower-albedo tree cover), thus raising temperatures 
For electricity generation, hydropower dominates RE sources, accounting for 15.6% of global total electricity in 2019. Only 32.7% of global hydro generated (and used) in that year was in Organization for Economic Cooperation and Development (OECD) countries, although the OECD produced most of the world’s wind, solar, and other non-hydro RE electricity 
. The gross global technical potential for hydro may be large 
, but the exploitable
potential (which considers environmental constraints) is far smaller, estimated at around 16,000 terawatt-hours (TWh = 1012
, compared with the 2019 hydropower output of 4222 TWh 
. However, even at this low present level, serious environmental problems are being encountered 
. Climate-induced changes in stream-flow temporal patterns, reservoir siltation and evaporation, and even Amazon deforestation, could all adversely affect levels of hydroelectricity in coming decades 
. Further, as Laghari 
has stressed, in important regions such as the Himalayas, shrinking glaciers from climate change could eventually lower the region’s hydropotential.
Output from the other main dispatchable RE electricity sources, bioenergy and geothermal, are negligible and only growing slowly 
. As already discussed, sustainable biomass potential for all energy uses, including electricity production, is limited. Global technical potential for geothermal electricity, estimated to be in the range 1.1–22 EJ, is also very small, and could never supply more than a small fraction of even the 2019 electrical output of 237 EJ 
. From around 2011, annual global growth in installed capacity has been linear, with about 250 megawatts (MW = 106
watt) added annually 
. Enhanced Geothermal Systems have been researched for more than two decades, but there is still no widespread utilization of such systems 
. Induced seismicity is a risk and has halted several projects.
The case of Japan, with a half-century of geothermal power development, is instructive. At an estimated 23,400 MW, Japan has the world’s third largest geothermal potential, after the USA and Indonesia 
. However, in 2019, only 525 MW was in operation, or a little over 2% of the potential 
. Yet the average capacity factor for geothermal power from its around 20 power plants has steadily fallen from over 80% in the early 1970s to under 60% by 2010, even given technology advances 
. This steady decline suggests that the feasible potential may be up to an order of magnitude lower than the published potential. Furthermore, around 80% of the published potential is located inside national parks. There is large global theoretical potential for geothermal heat
, but actual use will be restricted, because it can only be used locally 
The potential for tidal electricity is likewise small. According to van Haren 
, probably less than 100 gigawatts GW of tidal power could be installed globally, compared with about 623 GW for wind power already installed by 2019 
. He concluded that its small potential still had significant ecological costs. In 2019, total installed capacity was only 531 MW, almost entirely in France and South Korea 
The conclusion is that for a RE future, we must rely heavily on intermittent RE sources, chiefly wind and solar, a conclusion shared by both the US Energy Information Administration [EIA] 
and the IEA 
. These two energy sources would then need to meet nearly all global energy needs, both electric and non-electric 
. Even if wind and solar energy were not intermittent, some conversion to another energy carrier (possibly hydrogen) would still be needed to meet these non-electric energy demands. Energy losses are inevitably involved in conversion to a non-electric form for storage.
3. Evaluation Under Future Conditions is Needed for a 100% RE Scenario
In 2018, global consumption of primary energy (excluding non-commercial energy) was 583.9 exajoule EJ, up 1.3% from the previous year. Fossil fuel use accounted for 84.3% of this total and was still rising in absolute terms. (Emissions of CO2
from FF combustion and industry rose to 34.17 billion tonnes, a 0.5% increase on 2018 emissions 
). In 2019 solar and wind combined contributed only 8.0% of global electricity 
. What will global energy use be in 2050? As is clear from the forecasts made in 2019 for the year 2020, predicting the future, even a year ahead, can be a risky business 
Global long-range primary energy projections—all made before the coronavirus pandemic impacted the world in 2020—are available from such organisations as the EIA 
, BP 
, ExxonMobil 
, the IEA 
, the International Renewable Energy Agency (IRENA) 
, and the Organization of Petroleum Exporting Countries (OPEC) 
. All have forecast global energy use continuing to rise to some extent out to 2040 and even beyond, mainly driven by continued growth in non-OECD countries. The EIA 
, for example, in their Base Case, forecast a 2040 global primary energy use of 780 EJ, up from 607 EJ in 2015 (both figures including non-commercial energy). Increases are predicted for all energy types, including FFs. Non-OECD nations, which have slower uptake of RE 
, are then forecast to account for nearly two-thirds of total energy use. The single-value OPEC and ExxonMobil forecasts for 2040 are, respectively, 745 EJ and 678 EJ. BP use two scenarios, Evolving Transition, and a lower energy scenario, Rapid Transition. The forecasts for 2040, which exclude non-commercial energy, are 749 EJ and 686 EJ, respectively.
The IEA 
have assessed future energy use and greenhouse gas (GHG) emissions under two scenarios: The New Policies Scenario (NPS) and the Sustainable Development Scenario (SDS). NPS ‘Incorporates existing energy policies as well as an assessment of the results likely to stem from the implementation of announced policy intentions.’ Under the NPS scenario, primary energy use would increase significantly, to 700 EJ by 2040. SDS goes far further, with ‘an integrated approach to achieving internationally agreed objectives on climate change’. Global energy use at 570 EJ in 2040, would be slightly below its 2019 level. Nevertheless, 60% of primary energy in 2040 would still come from FFs, with their emissions reduced by either carbon capture and storage, or negative emissions technologies. Dynamic energy considerations also suggest that net RE output growth will be constrained 
, particularly if EROEI values for RE are low.
Regardless of future energy levels, all forecasts assumed that strong global economic growth would continue. Such growth would presumably lead to global increases in land use, particularly land for urban development, industry, and transport infrastructure; water for agriculture and cities; and various materials, including copper and elements with limited global resources, that are nevertheless vital for applications such as electronics, but also for PV cells and wind turbine motors. It is true that RE sources other than wind and solar will continue to have some share in future energy production, and in a few countries could even be dominant. However, for simplicity in the ‘what if’ thought experiment which follows, the intermittent primary electricity sources wind and solar (including solar thermal plants) are assumed to not only provide all energy for a still growing global market economy, but to compete for vital RE inputs including land, water, and materials, with the non-energy sectors of the global economy.
4. EROEI Evaluation in the Literature
Knowing the EROEI for wind and solar energy is vital for assessing whether they can together provide energy at levels as least as high as today, and many published papers have discussed their EROEI values. However, it is often very difficult to compare the findings of different researchers. This problem arises, not only because of different conditions in the analyses (rated capacity, year of installation, insolation or wind speed levels, type of PV cells, etc.), but also because different researchers have used different approaches for calculation (process, input–output, or hybrid method) and different boundaries for their analyses. There is also a need to factor in changes in energy inputs as technology develops; EROEI can be time-dependent, based on the maturity of the technology.
Given this confusion, Hall and colleagues 
and de Castro and Capellán-Pérez 
have attempted stress the importance of EROEI, and to put its calculation on a more consistent footing. Hall 
argued that EROEI was ‘a unifying principle for biology, economics, and sustainability’. In his book 
, he illustrated its connection to economics, pointing out that for US oil and gas, EROEI and costs are inversely related. Further, the EROEI for RE sources is less than often assumed 
, because of externalities.
According to de Castro and Capellán-Pérez 
, the most-used EROEIs are: ‘standard’ (measured at the ‘farm-gate’); final (at the consumer point-of-use); and ‘extended’ (which includes indirect investments as further inputs). They regarded the extended EROEI as being the most relevant. Hall et al. 
have argued that for functioning of an industrial society, as assumed here, an EROEI value of at least 3 is needed for both corn-based ethanol and oil, measured at the farmgate/mine mouth. Weißbach et al. 
found an ‘economic threshold’ for EREOI of 7, while Fizaine and Court 
have even argued that for the US, an EROEI of 11 was needed for economic growth to continue. Once a minimum value of EREOI is assumed, it is, in principle, possible to read off the total energy available from the EROEI vs. cumulative output curves for each RE type. Raugei 
has, however, warned that the use of single values for minimum EROEI is too simplistic.
5. Improving EROEI Estimates in a 100% Wind/Solar System
5.1. Declining Quality of Inputs
For both solar and wind energy, the theoretical
potential must first be reduced by omitting from consideration areas off-limits for energy generation (settlements, national parks, mountainous regions, polar regions, etc.) to obtain the geographical
potential. Next, suitable areas will have technical constraints on production because of inevitable losses in converting, for example, wind energy to electricity with a turbine, and also because in many locations the energy produced would not be economically feasible. This further reduction leaves the technical
. A production curve can, in principle, then be constructed, showing cumulative (global) technical potential as a function of EROEI. Historically, the EROEI has been found to also decline for fossil fuels as a function of cumulative output 
For wind, EREOI decline will mainly be the result of having to tap progressively lower wind speeds as prime locations are progressively taken up. Given how vast the solar resource is, insolation levels are not likely to be a global constraint. Instead, remoteness from load centres, and lack of both infrastructure and fresh water in desert areas is likely to greatly reduce the geographical potential. An important problem will be the low value of insolation and hence EROEI in many populated areas, particularly Europe and Japan. Land can also be a constraint, with the EU-27 calculated as needing 50% of its land area for a 100% solar energy system 
. Further, indirect evidence for some land constraints is the increasing popularity of water-based floating PV plants, especially in Japan 
Dupont and Jeanmart 
developed output curves for both solar and wind energy possible in the European Union (EU) as a function of EROEI and showed that present EU energy use has already exceeded potential EU RE production. Their EROEI values were presumably based on existing conditions, not on a 100% wind/solar system. Dupont et al. 
examined global solar potential, concluding: ‘The resulting constrained solar potential worldwide was estimated at 1098 exajoules per year, of which 98%, 75%, and only 15% can be extracted if the system needs to deliver an energy return on energy invested set at 5, 7.5, and 9, respectively. The resulting global solar potential is substantially lower than most previous estimates.’
An increasingly important component of input energy will be that needed to mine and process the materials needed for PV and wind turbine manufacture. In the expanded economy assumed here by 2050 or so, materials demand for the non-energy sectors of the economy will also expand. Wiedmann et al. 
have shown the close correlation between global GDP and materials consumption. Mining already consumes some 8–10% of global energy 
Indeed, Van den Bergh et al. 
have asked whether moving from fossil fuels to RE will merely replace one environmental problem (climate change) by another—mineral scarcity. They coined the term ‘environmental problem shifting’ for this possibility.
5.2. Uncounted Environmental Costs
As Hoegh-Guldberg et al. 
have pointed out, with around 1.0 °C post-industrial global temperature rise, the world is already experiencing serious adverse effects. Furthermore, adverse changes are not linear with temperature rise, as shown by comparing adverse changes experienced by the first and second 0.5 °C rises. Nevertheless, as discussed by Lade et al. 
, serious climate change is merely one of the environmental challenges Earth faces; others include modification of the global biogeochemical flows, ocean acidification, and biosphere integrity (oceans, fresh water, land). Even though compared with bioenergy or hydropower, wind and solar energy have fewer (uncounted) environmental costs, they are not negligible 
. Although wind turbines cause bat and bird deaths, and PV cell production results in toxic wastes, the main environmental costs arise from mining input materials.
Hadian and Madani 
developed a ‘relative aggregate footprint’ for assessing the environmental sustainability of various energy sources. They found that PV solar energy was only marginally more environmentally sustainable than natural gas. Rehbein et al. 
have further warned that future RE growth could have a big negative impact on many important locations for global biodiversity.
These costs include adverse effects on wildlife, especially for wind energy, where bird and bat deaths are a problem. In future, however, the most serious environmental costs will probably come from mining progressively lower-quality ores. If ore concentrations decrease by a factor of 10, wastes (and the environmental problems they create) can also be expected to rise by a factor of 10.
5.3. Need for Energy Conversion and Storage
EROEI values are often measured as the ratio of electricity output to primary energy inputs. For dispatchable electricity sources, this underestimates the real EROEI value. However, as already mentioned, given the assumption of a 100% wind/solar system, conversion and storage of at least some of the intermittent electricity output will be needed, because:
Output, even if constant, will usually not match instantaneous demand load.
Output will, in fact, vary over time, with periods of over- and under-production, even if the demand load were constant. Expanding the coverage of the solar/wind power grid will smooth output to some extent, but at the cost of extra transmission infrastructure.
Conversion and storage will still be needed even if, as expected, the share of electricity in final energy demand is higher than today. However, this conversion need comes at a heavy energy cost. Energy can be stored in batteries, compressed air, or pumped hydro schemes, but then cannot be directly used for non-electric purposes. In contrast, H2 storage is more versatile, as it can be used in stationary fuel cells to produce combined heat and power. According to Ajanovic and Haas 
, using hydrolysis to produce hydrogen (H2) for direct use as a fuel is about 50–60% energy efficient for the full cycle. Direct H2 use would be far more efficient than re-conversion of the H2 to electricity, with only about 27-38% full-cycle efficiency. There are also further input energy costs and losses associated with storage itself, for example as compressed H2 in tanks or underground caverns.
5.4. Removal of Fossil Fuel ‘Subsidy’
Some analyses have concluded that fossil fuels have much higher EROEI values than all renewables except hydropower 
. Brockway and colleagues 
found that published EROEI estimates for FF were typically 25 or more, if both FF output and input energy were assessed in thermal units. Raugei 
, in a response to Weißbach et al., claimed that the energy content of the FFs should be included as energy inputs. However, if at least some fossil fuel reserves remain unused to mitigate climate change, their energy content should be ignored, as it is for wind and solar energy. When the corrections discussed in the preceding three sub-sections are included, the EROEI difference is likely to be even greater. It might also be objected that the EROEI values calculated for FFs would be much lower if the energy costs for carbon capture and sequestration (CCS) were factored in 
. However, the EROEI values for FFs do not at present include these costs—and so, nor do present EROEI calculations for REs, with their FF inputs, discussed above.