The Feasibility of Powering the World with Wind, Water, and Solar Power

Jacobson and Delucchi (2011) address the pressing problem of climate change by proposing to produce all new power worldwide from wind, water and sunlight (WWS) by 2030 and to replace pre-existing energy sources with WWS by 2050. In this Part I of their two-part study, they assess the feasibility of doing so by calculating global end-use energy demand in a WWS world and comparing it to that of a world powered by fossil fuels as projected by the Energy Information Administration. They also examined worldwide capacity for WWS energy production and the limitations of materials used for the construction of WWS infrastructure. They estimated that about 3,800,000 5 MW wind turbines, 49,000 300 MW concentrated solar plants, 40,000 300 MW solar PV power plants, 1.7 billion 3kW rooftop PV systems, 5350 100 MW geothermal power plants, 270 new 1300 MW hydroelectric power plants, 720000 0.75 MW wave devices, and 490,000 1 MW tidal turbines can meet global energy demand in 2030 with a 1.0% increase in land use, and found that barriers are primarily social and political rather than technical or economic. —Lucinda Block
Jacobson, M. Z., Delucchi, M. A., 2011. Providing all global energy with wind, water and solar power, Part I: Technologies, energy resources, quantities and areas of infrastructure, and materials. Energy Policy 39, 1154–1169.

          Mark Z. Jacobson and Mark A. Delucchi found a decrease in global end-use energy demand in 2030 compared to the Energy Information Administration’s projections, which predicted demand will increase from 12.5 trillion watts (TW) to 17 TW in the year 2030 given an energy supply similar to today’s, constituted by 35% oil, 27% coal, 23% natural gas, 6% nuclear, and the rest from biomass, sunlight, wind, and geothermal. In the WWS scenario proposed by the authors, all end uses that can be electrified would use WWS power directly, and end uses that require combustion (like industrial processes) would use electrolytic hydrogen produced with WWS. Heating and cooling processes would employ electric heat pumps, and batteries, fuel cells, or a hybrid of the two would replace liquid fuels in non-aviation transportation. Aviation would use liquefied hydrogen to then be combusted. Jacobson and Delucchi calculated that an all WWS world would require approximately 30% less end-use power than the EIA projections for our current heavily fossil fuel-powered world. This is due to some increases in efficiency, for example, in the case of using electricity directly for heating or electric motors, as well as modest conservation measures (increases in efficiency through better insulation, more efficient lighting and heating, passive heating and cooling in buildings, and large-scale planning to reduce energy demand) and subtracting the energy requirements of petroleum refining.
          The authors investigated the availability of renewable resources that could potentially be exploited for power production in order to evaluate whether these could meet global energy demands in 2030. They found that wind and solar power in likely-developable locations could each provide enough power by themselves to meet global demands, with wind power potentially providing 3–5 times global demand and solar power potentially providing 15–20 times global demand. Concentrated solar power (CSP) could also meet global demand and has the ability to store energy for night usage, but it requires more land than PV and can use about 8 gal/kWh of water in a water-cooled plant, compared to almost no water for PV or wind. At the same time, air-cooled plants could be a viable alternative to water-cooled plants in areas of scarce water resources. Although other WWS technologies like wave power, geothermal and hydropower have much less energy potential (between 0.02 TW for tidal power and 1.6 TW for hydroelectric), Jacobson and Delucchi say they will be more abundant and economical than wind or solar in many locations and that since wind and solar power are variable, these other technologies could help stabilize electric power supply.
          In the WWS power generation scenario created by the authors, 50% of power will come from wind, 20% from CSP plants, 14% from solar PV plants, 6% from rooftop PV, 4% each from hydrothermal and geothermal plants, and 1% each from wave and tidal energy. Calculating combined footprint and spacing areas required for these technologies led Jacobson and Delucchi to the conclusion that their WWS scheme will require an additional 1.0% of global land area.
          The authors found that resource availability of bulk materials like steel and concrete is unlikely to constrain the development of WWS power systems. Some of the rarer materials used for WWS technologies include neodymium for electric motors and generators, platinum for fuel cells, and lithium in batteries, however, could present problems. Wind power is currently limited by neodymium requirements for permanent magnets in generators. Solar power is limited by silver reserves, although research suggests that opportunities exist to produce PV power with low cost and commonly available materials. Current neodymium requirements for electric motors similarly imply a need to develop alternative motors that do not use rare-earth elements. Global reserves of lithium are limited and in order to satisfy requirements for electric vehicles and other uses, a global recycling program is needed. Similarly, a platinum recycling program would be required in a scenario of producing 20 million hydrogen fuel cell vehicles per year, which could easily deplete platinum reserves in less than 100 years. Although Jacobson and Deluccchi expect the cost of recycling or replacing neodymium or platinum to be negligible, this is dependent on a drastic improvement in worldwide recycling infrastructure and in many cases finding viable alternatives to existing technologies.
          Jacobson and Delucchi find a world powered entirely by wind, water and solar power to be feasible, with a marked decrease in global end-use energy demand, a 1.0% increase in land use, and some need for technological substitutions and/or recycling programs for materials used in renewable energy construction. The authors recommend replacing all new energy with WWS by 2030 and all existing energy with WWS by 2050. The study does not provide a life-cycle analysis of implementation of the proposed WWS technologies. This would potentially be useful, as it would require a measured analysis of all environmental impacts, including impacts of natural resource extraction for new infrastructure. Interestingly, Jacobson and Delucchi neglect to consider factors like the CO2 emissions from the chemical process of making cement, which would be required on a large scale for the production of wind turbines in their scenario.
The authors have published Part II to this study, in which they consider reliability, system and transmission costs, and policies needed to implement worldwide WWS infrastructure. 

Estimating our Commitment to Global Warming

Transitioning from a fossil fuel-based economy to one based on renewable energy is impeded by widespread existing energy infrastructure: not only primary energy infrastructure such as coal<!–[if supportFields]> XE “coal” <![endif]–><!–[if supportFields]><![endif]–>-fired power plants, but also transportation infrastructure such as motor vehicles or airplanes and residential infrastructure such as natural gas-burning furnaces or stoves. Unless this infrastructure is prematurely decommissioned or widely retrofitted with expensive carbon capture and storage<!–[if supportFields]> XE “carbon capture and storage (CCS)” <![endif]–><!–[if supportFields]><![endif]–> technology, this “infrastructural inertia” represents committed CO2 emissions as we move into the future. Davis et al. (2010) calculated the cumulative future emissions of existing energy infrastructure and found that if we completely discontinued the production of net CO2-emitting infrastructure, existing infrastructure alone would contribute 496 gigatonnes of CO2 to the atmosphere between 2010 and 2060, increasing mean global temperatures by 1.3 °C. Noting the difference between this quantity and estimated future warming, the authors conclude that the sources of most emissions are yet to be built. However, they believe that extraordinary efforts are required to prevent the continued expansion of CO2-emitting infrastructure. —Lucinda Block

Davis, S. J., Caldeira, K., Matthews, D., 2010. Future CO2 emissions and climate change from existing energy infrastructure. Science 329, 1330–1333.

          Steven J. Davis and Ken Caldeira of the Carnegie Institution of Washington along with Damon Matthews of Concordia University in Montreal used datasets of worldwide CO2 emissions from directly emitting infrastructure such as power plants and motor vehicles as well as estimates of emissions produced by industry, households, businesses, and other forms of transport to predict cumulative global CO2 emissions through 2060. Historical data provided them with lifetimes and annual emissions of infrastructure. The authors estimated emissions from non-energy sources such as land use change or agriculture<!–[if supportFields]> XE “agriculture” <![endif]–><!–[if supportFields]><![endif]–> using the International Panel on Climate Change’s (IPCC<!–[if supportFields]> XE “Intergovernmental Panel on Climate Change (IPCC)” <![endif]–><!–[if supportFields]><![endif]–>) Special Report on Emissions Scenarios A2 scenario<!–[if supportFields]> XE “IPCC A2 scenario” <![endif]–><!–[if supportFields]><![endif]–>. They used an intermediate-complexity coupled climate-carbon model, the University of Victoria Earth System Climate Model, in order to calculate changes in atmospheric CO2 and temperature based on emissions.
          Davis et al. calculated a cumulative 496 gigatonnes (1 Gt=1012 kg) of global CO2 emissions between 2010 and 2060, with 282 and 701 Gt CO2 being the lower and upper bound estimates. Accounting for non-energy CO2 emissions, the total atmospheric CO2 in this scenario stabilizes below 430 parts per million (ppm), with an increase of global temperatures of 1.3 °C (1.1–1.4 °C above pre-industrial levels or 0.3–0.7 °C above current temperatures). The authors calculate emissions through 2060 (as opposed to through 2100, as with many other climate predictions) because by 2060 all energy-related sources of CO2 emissions are predicted to be no longer functional. Whereas they calculate a mean cumulative emissions of 496 Gt CO2 from existing energy infrastructure, scenarios considering the continued expansion of fossil fuel-based infrastructure through 2100 predict cumulative global emissions of 2986 to 7402 Gt CO2. In those scenarios, global temperatures increase by 2.4–4.6 °C above pre-industrial levels and atmospheric CO2 stabilizes above 600 ppm. Internationally, a rise in temperature of 2 °C and an atmospheric CO2 level of 450 ppm are considered to be the benchmark past which geophysical, biological and socioeconomic systems are especially vulnerable. Thus, the authors note, as existing energy infrastructure does not surpass the benchmark, the infrastructure that represents the most threatening CO2 emissions has yet to be built.
          Existing energy infrastructure is concentrated in highly developed countries such as Western Europe<!–[if supportFields]> XE “Europe” <![endif]–><!–[if supportFields]><![endif]–>, the United States, and Japan and populous countries experiencing rapid development, particularly China<!–[if supportFields]> XE “China” <![endif]–><!–[if supportFields]><![endif]–>. China accounts for the greatest energy inertia, where almost one quarter of worldwide electrical generating capacity has been commissioned as coal<!–[if supportFields]> XE “coal” <![endif]–><!–[if supportFields]><![endif]–> plants since 2000. The young age of its existing infrastructure compared to that of the U.S., Japan, or Western Europe also contributes to China’s large emissions commitment, approximately 37% of the global total. However, emissions commitment per capita in China is comparable to Japan and Western Europe and far less than that of the U.S. (136 tons CO2 per person versus 241). Davis et al. emphasize the importance of historic emissions in already developed countries and consumption in those countries as a driving force of Chinese emissions. They also note that committed emissions per unit of GDP is much higher in developing countries than already developed ones, showing that infrastructural inertia of emissions is greatest where industrialization is occurring but incomplete.
           Davis et al. conclude that although their estimates of cumulative committed global emissions of CO2 do not push us past the threshold of 450 ppm CO2 and 2 °C of warming, avoiding great quantities of CO2 emissions from not yet built infrastructure will require a tremendous political effort and shift, partially because of the supporting infrastructure for CO2 emitting devices such as highways or factories that produce internal combustion engines. Though their findings do not have groundbreaking implications for climate change studies, the study provides a useful benchmark of what future emissions are inevitable without high-cost retrofitting or halting of industry and what future emissions can more easily be reduced. 

Re-evaluating Lifecycle Greenhouse Gas Emissions of Natural Gas Production

Natural gas is widely regarded as a transitional or bridge fuel: though still a fossil fuel, it is believed to have less global warming potential<!–[if supportFields]> XE “global warming potential (GWP)” <![endif]–><!–[if supportFields]><![endif]–> (GWP) than oil or coal<!–[if supportFields]> XE “coal” <![endif]–><!–[if supportFields]><![endif]–>. In this paper Howarth et al. (2011) re-examine that assumption by calculating fugitive methane<!–[if supportFields]> XE “methane (CH4)” <![endif]–><!–[if supportFields]><![endif]–> emissions from both conventional and unconventional gas resources, namely shale gas from high-volume hydraulic fracturing<!–[if supportFields]> XE “hydraulic fracturing” <![endif]–><!–[if supportFields]><![endif]–>. Shale gas production has boomed in recent years, exceeding conventional production in 2009, and is expected to constitute a large part of the gas used to transition from fossil fuels to renewable energy. However, few scientists have evaluated the greenhouse gas footprint of unconventional gas, and in light of recent findings that methane has an even greater GWP than previously believed, it is imperative to re-assess the GWP of natural gas production from conventional and unconventional sources. Howarth et al. calculated that 3.6­–7.9% of methane from shale gas production escapes through fugitive emissions and venting, a percentage 1.3–2.1 times higher than from conventional gas production. They found shale gas to have a larger GWP than oil and coal on a 20-year timescale, and to have a larger GWP than oil and comparable GWP to coal on a 100-year timescale.—Lucinda Block
Howarth, R. W., Santoro, R., Ingraffea, A., 2011. Methane and the greenhouse-gas footprint of natural gas from shale formations. Climate Change Letters, forthcoming.

          Howarth et al. used two recently available reports for their data, a 2010 Environmental Protection Agency (EPA) document on greenhouse gas emissions from the oil and gas industry and a 2010 Government Accountability Office report on natural gas losses on federal lands. The former document is the first update on oil and gas emissions factors since 1996, when the agency produced a report that served as the basis for the national greenhouse gas inventory for the past decade. Howarth et al. remark that the 1996 study was neither based on random sampling nor comprehensive; instead, data were collected from model facilities through voluntary reporting. The EPA acknowledge in their new report that emissions factors are much higher from some sources than originally thought, and that the first report was published at a time when methane<!–[if supportFields]> XE “methane (CH4)” <![endif]–><!–[if supportFields]><![endif]–> emissions were not a significant concern.
          Because recent studies suggest that methane<!–[if supportFields]> XE “methane (CH4)” <![endif]–><!–[if supportFields]><![endif]–> has a greater GWP<!–[if supportFields]> XE “global warming potential (GWP)” <![endif]–><!–[if supportFields]><![endif]–> than previously thought—having 33 times the GWP of CO2 when examined on a 100 year timescale, whereas previously thought to be 25 (IPCC<!–[if supportFields]> XE “Intergovernmental Panel on Climate Change (IPCC)” <![endif]–><!–[if supportFields]><![endif]–> 2007)—and because of the presumably low methane emissions factors used in recent years to calculate greenhouse gas inventories, Howarth et al. focus in this paper on calculating the fugitive methane emissions and emissions from venting throughout conventional and unconventional natural gas production.
          The main differences the authors find in methane<!–[if supportFields]> XE “methane (CH4)” <![endif]–><!–[if supportFields]><![endif]–> emissions between conventional and unconventional gas production occurs during well completion. The extraction of shale gas requires high-volume hydraulic fracturing<!–[if supportFields]> XE “hydraulic fracturing” <![endif]–><!–[if supportFields]><![endif]–>, a process in which large volumes of water are pumped into wells in order to fracture and re-fracture the otherwise impermeable shale and stimulate gas flow. Much of this water returns to the surface as “flow-back,” carrying with it large quantities of methane. Howarth et al. used fairly uncertain data to calculate the flow-back emissions of five different unconventional formations, with some of it coming from PowerPoint slides of EPA-sponsored workshops. They took the mean of methane losses from flow-back as a percentage of total lifetime production of the well, resulting in a figure of 1.6%.
To calculate the total percentage of gas loss from well completion the authors had to add to this number the methane<!–[if supportFields]> XE “methane (CH4)” <![endif]–><!–[if supportFields]><![endif]–> lost from “drill-out,” the stage of high-volume hydraulic fracturing<!–[if supportFields]> XE “hydraulic fracturing” <![endif]–><!–[if supportFields]><![endif]–> in which producers drill out the plugs used to separate fracturing stages. Because drill-out emissions were not available for individual formations, Howarth et al. used the mean of EPA drill-out emissions estimates (142,000 to 425,000 m3 of methane; mean=280,000 m3) multiplied by the average lifetime production of four of the rock formations used to determine flow-back emissions. Because of the higher lifetime production of these formations compared to others examined in a study of twelve formations, the calculation resulted in a conservative estimate for drill-out methane emissions as a percentage of gross well production of 0.33%. Combined with flow-back emissions, Howarth et al. calculate that 1.9% of gross shale gas production is emitted during well completion as an uncertain but highly conservative estimate. This figure contrasts with an emissions estimate of 0.01% of gross gas production for conventional wells using EPA data.
          Howarth et al. calculate a series of ranges for other potential sources of fugitive methane<!–[if supportFields]> XE “methane (CH4)” <![endif]–><!–[if supportFields]><![endif]–> emissions in conventional and shale gas production. The authors attribute the same methane loss ranges to both conventional and unconventional gas production for equipment leaks and routine venting, since the same technologies are used for both types of gas once they are connected to a pipeline. The low end of 0.3% represents the use of best available technology, and the upper estimate of 1.9% does not include accidents or emergency venting.
Fugitive emissions from transport, storage, and distribution of natural gas can also be assumed to be the same for both types of gas production, as conventional and shale gas are an identical commodity at this point of production. The authors employed the figure of 1.4%, calculated in a 2005 study by Lelieveld et al. as the lower estimate of average gas loss combining transport, storage, and distribution losses. They found the estimate’s upper limit using a bottom-down approach, calculating the disparity between measured gas produced at the wellhead and measured volume of gas purchased and consumed as an end product. The authors calculated an upper limit of 3.6% by taking the mean of the State of Texas’s data for missing and unaccounted gas in the years 2000 and 2007. Howarth et al. believe 3.6% to still be a conservative estimate, given that industry fought a proposed hard cap on missing and unaccounted gas of 5% in the state.
The authors also provided a range of methane<!–[if supportFields]> XE “methane (CH4)” <![endif]–><!–[if supportFields]><![endif]–> losses for “liquid unloading” and gas processing. Liquid unloading is a process often needed for conventional well production and sometimes for unconventional well production, in which liquid is unloaded to mitigate water intrusion as reservoir pressure drops when a well matures. Though methane losses from liquid unloading were estimated at 0.02–0.26% of gross production, because some wells do not require liquid unloading, the authors used the range of 0–0.26%. Surprisingly, they used the same range for both conventional and shale gas, despite acknowledging that liquid unloading is primarily required for conventional wells. Logically, conventional gas should have been given a higher estimate for liquid unloading than shale gas, but the authors do not address this.
Similarly, the authors provided the same range for both conventional and shale gas for emissions from gas processing. Howarth et al. explain that both conventional and shale gas vary in quality when extracted, and thus sometimes require processing that creates more methane<!–[if supportFields]> XE “methane (CH4)” <![endif]–><!–[if supportFields]><![endif]–> emissions. However, they do not address the question of whether one type of gas or the other has a higher average quality when extracted. Instead, the authors provided a methane loss range of 0%, representing no processing, to 0.19% of production for both types of gas. This area of uncertainty goes unaddressed in the study.
Howarth et al. calculated the total fugitive methane<!–[if supportFields]> XE “methane (CH4)” <![endif]–><!–[if supportFields]><![endif]–> loss as a percentage of gross production at 1.7–6.0% for conventional gas, and 3.6–7.9% for shale gas. Compared to coal<!–[if supportFields]> XE “coal” <![endif]–><!–[if supportFields]><![endif]–> (both surface- and deep-mined) and diesel oil on a 20- and 100-year timescale, the authors found the GWP<!–[if supportFields]> XE “global warming potential (GWP)” <![endif]–><!–[if supportFields]><![endif]–> of shale gas to be 1.2–2.1 times greater than coal and 1.5–2.5 times greater than oil on a 20-year timescale. On a 100-year timescale, they found the GWP of shale gas to be comparable to coal and up to 1.4 times greater than oil. The high estimate GWP of conventional gas was also significantly higher than oil and coal on a 20-year timescale and comparable on a 100-year timescale.
Comparing their estimates for conventional gas fugitive methane<!–[if supportFields]> XE “methane (CH4)” <![endif]–><!–[if supportFields]><![endif]–> emissions to other peer-reviewed literature, the authors note that although two of three studies found lower estimates for fugitive emissions, these studies used GWP<!–[if supportFields]> XE “global warming potential (GWP)” <![endif]–><!–[if supportFields]><![endif]–> factors for methane that are now known to be too low and still concluded that in many cases a switch to natural gas from coal<!–[if supportFields]> XE “coal” <![endif]–><!–[if supportFields]><![endif]–> could aggravate rather than mitigate the effects of climate change. Lelieveld et al. concluded that natural gas would be worse than oil if fugitive methane emissions exceeded 3.1% of total production, and worse than coal if they exceeded 5.6%. Adjusting that study’s GWP factor for methane to account for recent findings, Howarth et al. claim that fugitive methane emissions of only 2–3% make natural gas more impactful than oil and coal, well within both their emissions ranges for conventional and shale gas.

          In conclusion, the authors emphasize that rather than promoting continued use of oil and coal<!–[if supportFields]> XE “coal” <![endif]–><!–[if supportFields]><![endif]–>, they warn against policymaking that relies on natural gas as a bridge fuel and assumes that gas implies lower carbon emissions per unit of energy produced compared to other fuels. However, they acknowledge that there is a large amount of uncertainty in fugitive methane<!–[if supportFields]> XE “methane (CH4)” <![endif]–><!–[if supportFields]><![endif]–> estimates and recommend further study, given the importance of the topic.

Evaluating the Uncertainty in Calculating Greenhouse Gas Emissions for Electricity Generation

Because 40% of U.S. CO2 emissions come from electricity generation and distribution, the ability to calculate CO2 emissions per unit of electricity consumed is crucial in order to perform a life-cycle analysis (LCA), be it of a product or process.  However, the greenhouse gas emissions associated with an individual entity’s electricity consumption is nearly impossible to calculate given the nature of electricity grids.  For this reason, LCA practitioners often employ emissions factors, or estimated average quantity of CO2 emitted per unit of energy consumed.  Unfortunately, emissions factors vary greatly both spatially and temporally due to different energy sources used for generation, as well as differing plant efficiencies. The authors point out that in addition to electricity coming from varying sources (for example, hydroelectric power provides much of the Pacific Northwests’s electricity due to the natural availability of that resource), electricity systems are quite complex because deregulation in the 1990s connected more remote customers with more remote generators, making it even more difficult to trace the source and associated greenhouse gas emissions of one’s electricity.  In this study Weber et al. (2010) calculated the variability in emissions factor estimates and demonstrated the uncertainty in using these estimates for LCA and policymaking.  The authors also made suggestions for how to deal with this uncertainty.—Lucy Block
Weber, C., Jaramillo, P., Marriott, J., and Samaras, C., 2010. Life Cycle Assessment and Grid Electricity: What Do We Know and What Can We Know? Environmental Science & Technology 44, 1895-1901.

          Christopher Weber, Paulina Jaramillo, Joe Marriott, and Constantine Samaras examine the uncertainty of emissions factors at various geographic levels of the U.S. and in different locales by collecting different emissions factors for CO2, SO2 and NOx (though CO2 contributes primarily to global warming and is thus the main focus of the paper).  The authors acknowledge that they did not take into account the emissions of upstream supply chains for electricity generation, noting that accounting for upstream emissions would only slightly increase uncertainty.  The authors calculated emissions factors along several potential regional delineations of the electric grid.  The emissions factor with the largest geographical area was the U.S. continental average (0.69 kg CO2/kWh), followed by three regions based on electrical grid connectivity—the Eastern, Western, and Texas Interconnects.  At a smaller level, Weber et al. used the 24 subregional grid delineations as defined by the EPA’s eGrid and used in the Greenhouse Gas Protocol, a tool for conducting LCAs.  Finally, the authors used data collected by the U.S. Energy Information Administration through voluntary greenhouse gas reporting since 1992.  The different datasets considered form seven independent estimates of electricity emission factors for every combination of U.S. state, eGrid subregion, and grid operator (whether independent system operators or regional transmission operators). 
For their dataset, the authors calculated a coefficient of variation (COV), or the normalized standard deviation.  A higher COV meant more variation between different estimates for electricity emissions factor, and therefore a higher uncertainty of amount of CO2 emitted per unit of electricity generation in the region.  The average CO2, COV for all delineations, or districts, considered out of 101 total was 0.19 (an average uncertainty of ±40% at two standard deviations) and ranged from a maximum of 0.70 to a minimum of 0.08.  The districts with highest associated uncertainty were those that had smaller or larger than average local or regional emissions factors.  Since electricity grids do not correlate closely with state borders, emissions factors estimated along state lines had higher variation than those estimated according to eGrid delineations. 
The authors conclude that LCA practitioners and policymakers generally do not have access to the data required in order to calculate a specific consumer’s electricity-related greenhouse gas emissions.  Therefore, for practical purposes, Weber et al. recommend that standards organizations provide clear guidelines for conducting LCA calculations, and by standardizing these calculations reduce overall comparative uncertainty between different LCAs.  The authors suggest that standards organizations should discourage the use of political borders in calculating emissions intensity for a particular area, as this unnecessarily increases uncertainty.  Furthermore, researchers should report kWhs consumed alongside the assumed grid emissions factor within an appropriate electricity system delineation, in order to increase transparency and allow for normalized comparisons of a specific product.  If estimating indirect CO2 emissions is required, Weber et al. suggest that researchers provide a range for the emissions factor.  In that case, if an entity wants to guarantee an emissions reduction or carbon neutrality, it can use the highest range of emissions factors. 
In public policy decisions, choosing a set of emissions factors will raise issues of equity.  If too general a set of emissions were to be used and an emissions trading market were to be set up, local distribution companies buying lower-carbon electricity would obtain an advantage, and local distribution companies buying higher-carbon electricity would be at a disadvantage.  Additionally, using more locally specific emissions factors could potentially penalize energy users in areas that have higher-carbon electricity simply due to natural resources.  For example, electricity in the Pacific Northwest will be lower-carbon because of the regional hydroelectric resources.  An industry located in the Pacific Northwest stands to lose less from policies to reduce carbon emissions than industries in other regions. 
The authors note that while it may be possible, depending on required level of accuracy for the investigation, to choose an appropriate emissions factor (e.g., if an industry operates in many locales throughout the country and the investigation does not require a particularly high level of accuracy in emissions calculations, one could use the national average emissions factor), consistency in calculating the indirect emissions of electricity consumption is of highest importance, along with transparency and reproducibility of methods.  

Photovoltaic Technology in Regions of Low Solar Irradiation: A Broad Assessment of Environmental Impact

Whereas most lifecycle assessments (LCAs) use one-dimensional indicators and only apply to areas of high solar irradiation, Laleman et al. (2011) used both one-dimensional indicators and the multi-dimensional Eco-Indicator 99 (EI 99) to conduct a broad assessment the environmental impact of various photovoltaic (PV) technologies employed in areas of low solar irradiation such as Canada and Northern Europe.  Furthermore, they used these same indicators to compare PV systems to other sources of electricity production.  The authors found the energy payback time of PV systems to be less than 5 years, and the global warming potential to be approximately 10 times lower than a coal plant and 4 times higher than a nuclear power plant or wind farm.  The authors obtained significantly different results using EI 99 compared to one-dimensional indicators, and thus stressed the importance of carefully evaluating a combination of different environmental impact assessment approaches.—Lucy Block
Laleman, R., Albrecht, J., and Dewulf, J., 2011. Life Cycle Analysis to estimate impact of residential photovoltaic systems in regions with a lower solar irradiation. Renewable and Sustainable Energy Reviews 15, 267-281.

          Ruben Laleman, Johan Alrecht, and Jo Dewulf of Ghent University in Belgium used lifecycle data from the Ecoinvent database (v2.0) to assess the environmental impact of six different PV technologies under conditions of low solar irradiation (900-1000 kWh/m2/year).  As opposed to only using a one-dimensional indicator such as Cumulative Energy Demand (CED), Energy Payback Time (EPT), or Global Warming Potential (GWP), as many other authors conducting LCAs do, Laleman et al. compared environmental impact findings of these one-dimensional indicators to the multi-dimensional EI 99.  The authors also compared their findings of PV environmental impacts to the impact of other electricity sources such as hard coal, natural gas, and the Belgian electricity mix.
          The authors’ findings of PV technology’s environmental impact for the one-dimensional indicators—CED, EPT, and GWP—were comparable to previous literature conducted on the subject, but their findings for the EI 99 had very little correlation with the one-dimensional indicators (at most 22%).  Therefore, they stress the importance of employing a multi-dimensional indicator, especially alongside one-dimensional indicators, in order to give the most nuanced picture possible of environmental impacts.
          Besides assessing environmental impact for various environmental indicators—mineral extraction, fossil fuels, respiratory effects, ozone layer  depletion, ionizing radiation, climate change, carcinogenics, land occupation, ecotoxicity, and acidification and eutrophication—EI 99 categorizes those indicators into three main dimensions: human health, ecosystem quality, and the depletion of non-renewable resources, and creates three different “perspectives”—i.e., three different ways to deal with the subjective process of weighting and normalizing results based on different rankings of preferences, values, and attitudes.  The three perspectives are Hierarchist, Egalitarian, and Individualist.  The Hierarchist represents the view of the “average scientist” who is presumed to follow the IPCC’s (International Panel on Climate Change) assessment reports on the effects of climate change, balance short- and long-term concerns, and bases her views on consensus.  The Egalitarian greatly values ecosystem quality, considers the very long term—another way of saying she is concerned with sustainability—and is highly risk-averse, potentially resulting in overestimation of risks.  The Egalitarian is prone to consider all possible negative environmental effects of a phenomenon like climate change as definite.  This view contrasts with that of the Individualist, who only considers “proven” effects (as opposed to effects based on consensus but around which there remains some doubt).  The Individualist does not place any importance in fossil fuel depletion; rather, she only considers the depletion of minerals relevant.  Furthermore, the Individualist’s perspective lies within a short-term time frame, whereas the Egalitarian thinks in terms of a very long time frame.  Laleman et al. emphasize the need to clarify and outline these different perspectives in LCAs employing EI 99 so as not to cause serious misinterpretations, and for clarity’s sake they also include unweighted results.
          First, the authors evaluated environmental impact using one-dimensional indicators for the following six PV technologies: Cadmium Telluride (CdTe), CuInSe2 (CIS), ribbon Si, multi crystalline Si (multi c-Si), mono crystalline Si (mono c-Si) and amorphous (a-Si).  The newer technologies are the CdTe, CIS and ribbon Si.  Using the same figure for yearly energy output and the same conversion coefficient for electricity generation efficiency, the authors’ calculations for CED and EPT indicators are proportional to one another.  Whereas CED measures total energy required to construct the PV system over its lifetime, EPT measures the amount of time until the PV system produces more energy than was required for its construction.  In these analyses, the newer technologies were found to be more efficient than older ones, requiring less than 30,000 megajoules equivalent per kilowatt-peak (kilowatt-peak [kWp] is a measure of solar energy output under laboratory conditions; a standard home installation is considered to be 3 kWp in this study) for their construction.  All PV types had an EPT of less than 5 years in low irradiation conditions.  CdTe, CIS and ribbon Si EPTs were about one year less than those of the other PV systems, though this difference decreased as irradiation conditions increased.  In high solar irradiation regions like Spain, EPTs were only 2­–3 years.
          The GWP measures quantity of greenhouse gases emitted over the lifecycle of a PV system.  As with CED and EPT indicators, the GWP indicator showed the three newer PV technologies, along with multi c-Si, to have less impact than the three older ones (approximately 5000 kg of CO2 equivalent compared to approximately 6000 kg of CO2 equivalent).
          The EI 99 results differed significantly from the one-dimensional indicators.  Using the Hierarchist perspective, CdTe was found to have the highest impact score, and greatly exceeded the scores of the other newer technologies (450 compared to 317 and 353).  A breakdown of impact scores according to individual environmental indicators shows that most impact originates from fossil fuels and respiratory effects.  The authors note that reducing the energy input of PV production will decrease the impact related to fossil fuel extraction, respiratory effects, climate change, acidification and carcinogenics as they all relate to one another. 
          In order to compare the environmental impact of PV technology to other sources of electricity, the authors selected the multi c-Si system, as it has the largest market share.  They employed both a pessimistic (20 year) and optimistic (30 year) lifespan estimate for the PV system, and using both GWP and EI 99 indicators they compared the impact for 1 kWh (kilowatt-hour) produced by the various electricity sources.
          The GWP analysis showed PV electricity to have a markedly lower impact than fossil fuel based sources (even with an expected lifespan of 20 years, the PV’s GWP was 0.12 kg of CO2 equivalent per kWh (kgCO2-eq/kWh) compared to 0.53 for natural gas).  The Belgian mix is surprisingly low, at 0.33 kgCO2-eq/kWh, due to the high (55%) proportion of nuclear energy contribution.  The GWP of PV electricity was found to be approximately four times higher than nuclear and wind and ten times lower than coal (the authors claim their impact assessment for nuclear takes into account the impact of radiation on human health).
          The EI 99 results for compared environmental impact across electricity sources varied greatly depending on the perspective used.  Because the Individualist perspective does not “value” fossil fuel extraction as having an environmental impact, the mineral extraction associated with PV construction is weighted very highly and thus the total impact of PV is very high for the Individualist compared to the Egalitarian and Hierarchist perspectives.  Since PV technology requires a significant level of aluminum, iron, and copper, the Individualist finds PV to be much more impactful than natural gas, whereas the Egalitarian and Hierarchist find natural gas to be significantly more impactful.  In the unweighted category of ecosystem quality, PV is about twice as impactful as natural gas (and both are small compared to coal).  In the category of human health and resource depletion, PV impacts are negligible, natural gas impacts are small, and coal impacts are high.  Though a comparison of mineral ore extraction across electricity sources show that PV requires a relatively large amount of mineral ore, an EI 99 assessment of overall resource depletion shows mineral extraction associated with PV to be negligible compared to the fossil fuel extraction required for other electricity sources.  With regards to the issue of mineral ore required for PV construction, the authors indicate that the removal of the aluminum frame used for PV panel installations would greatly reduce overall environmental impact, and they recommend an efficient recycling program for the ores. 

          The authors conclude that PV systems have a relatively low  environmental impact even in areas of low solar irradiation, especially compared to fossil fuel based sources of electricity, though mineral extraction requirements should be taken into consideration.  Lifetime energy production ranged from 4-6 times lifetime energy consumption, and could reach 12 times lifetime energy consumption in sunny regions.  Lifetime greenhouse gas emissions were significantly lower than fossil fuel based sources of electricity production.  The EI 99 analysis showed that when fossil fuels were considered to have a negative impact on the wellbeing of future generations, PV systems were found to be less impactful than natural gas, coal, and the Belgian electricity mix.  The weighting step of EI 99 analysis greatly affected results, making the Individualist perspective consider PV more impactful than natural gas—as the authors point out, many would consider the large weight the Individualist assigns to mineral extraction to be illogical or irrational in this case.  The authors suggest this implies a need for great care and consideration of complexities in conducting a LCA.  Furthermore, due to low correlation of EI 99 results with one-dimensional indicators, Laleman et al. recommend the use of various indicators for a thorough and comprehensive LCA.

A Baseline Life-Cycle Assessment for Hydrokinetic Energy Extraction

Miller et al. (2011) used life-cycle assessment methods to evaluate environmental impacts of hydrokinetic energy extraction (HEE), which have not previously been quantified.  The authors established a baseline methodology for doing so, and compared their LCA findings for HEE with other energy systems across a variety of environmental indicators.  HEE is considered an environmentally benign form of renewable energy, and by harnessing kinetic as opposed to potential energy, avoids the sediment movement and impact to ecosystem of conventional hydroelectric power generation.  HEE collects energy from the kinetic movement of water without constructing a dam, and therefore avoids large ecosystem disruptions and changes to water flow and volume associated with hydropower electricity generation.  In this study, the authors focus on HEE from rivers, as opposed to tidal flow or waves. They focus on the Gorlov system, which uses a helical crossflow turbine. Miller et al. found the Gorlov system compared closely to small hydropower and to have the lowest life-cycle impact of all energy systems considered.—Lucy Block

Miller, V., Landis, A., and Schaefer., 2011. A benchmark for life cycle air emissions and life cycle impact assessment of hydrokinetic energy extraction using life cycle assessment. Renewable Energy 36, 1040-1046.

          Veronica B. Miller, Amy E. Landis, and Laura A. Schaefer of the University of Pittsburgh employed the Tool for the Reduction and Assessment of Chemical and other environmental Impacts (TRACI) to conduct their Life-Cycle Impact Assessment (LCIA).  They considered upstream and downstream impacts of HEE, small hydropower, coal, natural gas, and nuclear power.  The small hydropower considerations included the dam structure, tunnel, turbine, generator, plant operation, and dismantling.  The calculation of coal plant impact included the coal production and preparation, coal processing, storage and transportation, and the calculation of natural gas impact included gas field exploration, natural gas production, gas purification, long distance transportation, and regional distribution.[1] The HEE LCIA included raw materials used to construct the Gorlov system, transportation, assembly, operation, and decommissioning. 
          Miller et al. compared different energy systems to each other across many environmental indicators to calculate which energy source implies the most environmental impact, and more specifically, how Gorlov HEE and hydropower compare to one another, but they do not mention how they controlled for the variable of energy output between different types of electricity plants.  The authors found that among all energy systems considered, coal and gas power plants had the highest environmental impacts related to global warming, acidification, eutrophication, ecosystem disruption, and smog formation.  Nuclear power was most the most impactful energy system in terms of ozone depletion.  Gorlov HEE and small hydropower had negligible global warming impacts compared to the other energy systems.  A comparison of Gorlov HEE and small hydropower showed that small hydropower was more impactful than Gorlov HEE across all indicators but respiratory effects and acidification.  The authors attribute these impacts to the production of copper, which is used for the Gorlov HEE generator, though they mention that methods of SO2-free copper production are currently being investigated.  The ecotoxicity of Gorlov HEE was shown to be less than 2% of that of small hydropower. 

          The study shows that Gorlov HEE is less impactful across almost all environmental indicators than small hydropower.  Nonetheless, the study should be viewed as a rough basis for estimating the life-cycle environmental impact of HEE. The data used for this study came primarily from estimates, as opposed to from a case study.  Additionally, many considerations were not taken into account: fish and local river ecology health was not taken into account, though Miller et al. suggest that a new LCIA category could be created and this impact could be calculated using estimated fish passage from HEE computational fluid dynamics and fish swimming data.  The authors did not consider in their assessment the variety of fiberglass types used for turbine blade construction.  Furthermore, the study did not account for negative environmental effects of dam construction, such as changes to overall water flow and temperature differences, or the benefits of hydropower, such as its reliability as a renewable source of energy and the benefit of creating a reservoir.


[1] The authors neglect to mention whether their data for the impacts of natural gas production have been updated to reflect recent developments in natural gas extraction from the use of horizontal hydraulic fracturing. Nonetheless, the impacts of this technology have not yet been thoroughly studied, so it is safe to assume they have not. 

The Environmental Impact of Electricity Production from Biomass: A Comparison of Poplar and Ethiopian Mustard

Spain has set the goal of producing 12% of its primary energy demand through renewable sources in its Renewable Energies Plan 2000–2010, and power generation from biomass represents an important contribution to meet this goal. Butner et al. (2010) evaluated the environmental performance of two biocrops used for electricity production, poplar and Ethiopian mustard.  Using a Life-Cycle Assessment (LCA) approach, the authors calculated environmental impacts for electricity generation from the two crops and compared values to natural gas generation and the Spanish electricity mix in order to assess whether or not these crops are environmentally competitive with conventional sources of power.  They found poplar to be less environmentally impactful than Ethiopian mustard, mostly due to higher production yields. Electricity from biomass had more impact in three of six environmental categories than natural gas power and less impact in all categories compared to the mix of electricity supplied to the Spanish grid.—Lucy Block
Butnar, I., Rodrigo, J., Gasol, C. M., and Castells, F., 2010. Life-cycle assessment of electricity from biomass: Case studies of two biocrops in Spain. Biomass and Bioenergy 34, 1780–1788.

          Isabela Butnar from the Universitat Rouvira I Virgili along with Julio Rodrigo, Carles M. Gasol, and Francesc Castells selected poplar and Ethiopian mustard for the study due to their high yields in Spain. They chose poplar in particular because of its strong environmental performance and high yields in Mediterranean areas compared with annual herbaceous crops, though its cultivation also consumes large amounts of water.  The two crops also differ significantly from one another: poplar is a perennial crop with a sixteen-year cultivation period, and Ethiopian mustard an annual herbaceous species.

The authors considered cradle-to-grave impacts of the different processes required to produce electricity from poplar and Ethiopian mustard, including field work, the use of farm machines, and the transport of materials associated with production such as fertilizer, herbicides, and packaging, as well as the transport of produced biomass to the power plant.  To calculate this last impact, Butnar et al. considered two different distances, 25 km and 50 km.  The authors based these distances on the percentage of available land for biocrop production at different plant capacities: when the required cultivated area of biomass for a given plant capacity exceeded 15% of the regional irrigated arable land, they calculated a distance of 50 km from the field; otherwise, they calculated a distance of 25 km. 
Along with distance values, Butnar et al. varied the power plant capacities and productivity yield values to calculate the environmental impacts of twelve different scenarios.  The three power plant capacities considered were 10, 25, and 50 MW.  The authors used two different productivity values for each crop, the lower limit of their productivity yield range and the average value.  For poplar these values were 9 and 13.5 t/ha, respectively, and for Ethiopian mustard they were 4.72 and 8.07 t/ha.
          Using SiAGROSOST, a software tool created by their research group, the authors found optimum values for minimizing environmental impact in the variety of scenarios mentioned above for ten different environmental indicators, though only six indicators were included in the report: acidification, global warming, human toxicity, ozone layer depletion, abiotic depletion, and photochemical oxidation.
          As expected, for all indicators, environmental impact decreased when biomass productivity increased.  Because of its higher productivity per hectare, poplar outperformed Ethiopian mustard, having less environmental impact across all indicators.  In general, a greater distance between fields and power plants (50 km as opposed to 25 km) implied greater environmental impact.  In turn, this had implications for the optimal power plant size—the larger capacity of 50 MW plants required more biomass cultivated area than 15% of the regional irrigated arable land, and thus the quantity of biomass required to run the plant at full capacity was not available at a 25 km distance.  This increased the environmental impact of 50 MW plants, despite their generally higher efficiency in energy production. 
          The two crops performed differently in their relationship with productivity, transport, and environmental impacts.  Poplar’s impacts were more closely associated with transport, and Ethiopian mustard’s with productivity.  When transport distances increased, poplar’s environmental impact increased more than Ethiopian mustard’s environmental impact did.  When productivity decreased, Ethiopian mustard’s environmental impact increased more than poplar’s did.  This information has implications for planning the use of different types of crops for biomass electricity generation: while both productivity and transport distance are important, either factor may be more important for different crops.
          By calculating the contribution of individual fieldwork activities—between fertilizers, harvesting, pesticides, and others—to the impact of crop cultivation, the authors found that fertilization is by far the most impactful step of cultivation (accounting for up to 78% of acidification and up to 82% of global warming impact).  However, replacing mineral fertilizers with natural fertilizers such as livestock manure could significantly reduce the environmental impact of fertilization, which, along with soil characteristics and weather, greatly contributes to crop productivity. 
          Butnar et al. compared their calculated impact values for biomass with electricity production from natural gas and the electricity put into the Spanish grid, which is largely dependent on fossil fuels.  They found that biomass electricity had a worse environmental profile than natural gas in the areas of acidification, human toxicity, and photochemical oxidation, and a better environmental profile in the areas of global warming, abiotic depletion, and ozone depletion.  Biomass electricity had a better environmental profile than the Spanish electricity mix in all areas.  However, it is important to note that Butnar et al. did not calculate the impact of transport and distribution of electricity in their biomass LCA, while values for natural gas and the Spanish mix include transport and distribution.  Additionally, Butnar et al. failed to account for the disposal of ashes from the combustion of biomass in their LCA.

          Butnar et al. found that electricity generation from poplar is more environmentally competitive than from Ethiopian mustard, and that power plants of 10 and 25 MW are more environmentally competitive for biomass electricity generation than 50 MW plants in the region of Tarragonès.  The authors recommend thoughtful biomass management plans that include recycled residues, e.g. from the cleaning of public parks.  They also state that in order to keep environmental impact of electricity generation from biomass low, biomass productivity must be optimized, and distances between field and power plant must be minimized.