14.4 — Energy, Source by Source

Deaths per unit of electricity generated, from all causes including accidents, mining, construction and air pollution, in deaths per terawatt-hour:

Source	Deaths per TWh (approximate)
Coal	~25
Oil	~18
Biomass	~4.6
Gas	~2.8
Hydropower	~1.3
Wind	~0.04
Nuclear	~0.03
Solar	~0.02

Nuclear power is, on the evidence, among the safest sources of energy ever used, and it is the one most people are most afraid of.

That gap between the measured risk and the perceived risk is a large part of why energy policy is difficult, and this chapter assesses each source on what it actually does.

The scale of the problem

Global energy consumption is around 180,000 terawatt-hours a year in primary energy terms.

Roughly 80 percent of it is fossil fuels, a share that has fallen only slowly despite enormous renewable growth, because total demand has grown at the same time.

And electricity is only about a fifth of final energy use. The rest is heat, industry and transport, several parts of which are far harder to decarbonise than power generation.

This matters because the public conversation is almost entirely about electricity, which is the part where the solutions are clearest and which is a minority of the problem.

Fossil fuels

Coal. The highest carbon intensity per unit of energy, the worst air pollution, and the largest death toll. Air pollution from fossil fuel combustion is estimated to cause several million premature deaths a year worldwide — a figure that dwarfs every industrial accident in history and that receives almost no attention because the deaths are dispersed and attributed to heart and lung disease.

And it is cheap where it is local, it is dispatchable — available on demand — and it supports large employment. India generates a majority of its electricity from coal and has substantial domestic reserves (Chapter 1.12), which makes it an energy security asset as well as a climate liability.

Oil. Principally transport, which is the hardest sector to substitute. Geopolitically concentrated (Chapter 12.5).

Gas. Roughly half the carbon dioxide of coal per unit of electricity and far less local air pollution, which makes it a genuine improvement.

And the methane leakage question is real and unresolved. Methane is a far more potent greenhouse gas over short periods (Chapter 14.1). Satellite measurement has found leakage rates from extraction and distribution substantially higher than official inventories in several regions. Above a leakage rate of roughly 3 percent, gas loses much of its advantage over coal on a twenty-year basis. Measured rates vary enormously by field and operator, which means the answer depends on which gas.

Nuclear

The evidence, stated plainly, because this is the source where public perception and measured outcome diverge most.

Chernobyl, 1986. The direct death toll from acute radiation was 28 to around 50. The UN scientific committee attributes around 5,000 thyroid cancer cases among people exposed as children, of which a small number were fatal — most were treatable and were caused by the failure to distribute iodine tablets and to restrict contaminated milk. Projections of eventual excess cancer deaths vary widely, from a few thousand to tens of thousands depending on the model used for low-dose exposure, and are not directly measurable against background rates.

Fukushima, 2011. The earthquake and tsunami killed around 18,000 people. The radiation released has been attributed one confirmed death from lung cancer in a worker. Over 2,000 deaths have been attributed to the evacuation itself — the disruption to elderly and hospitalised populations — which is a finding that has changed thinking about evacuation protocols.

Three Mile Island, 1979. No detectable health effects.

And the reaction was larger than the event. Germany accelerated its nuclear phase-out after Fukushima, and the shortfall was substantially met by coal in the following years, which on the table above means more deaths and more emissions.

Nuclear's genuine problems.

Cost and construction time. Recent Western projects have run enormously over budget and schedule — Hinkley Point C, Flamanville, Vogtle. The reasons include loss of construction expertise, first-of-a-kind designs, and regulatory change during construction. South Korea and China have built more cheaply and faster, which suggests the cost is not inherent.

Waste. High-level waste requires isolation for tens of thousands of years. The volume is small — the entire spent fuel output of a country's nuclear programme fits in a modest area — and no country except Finland has an operating deep geological repository. The problem is political rather than technical: nobody wants it in their constituency.

Proliferation. Enrichment and reprocessing technology has military applications (Chapter 11.10).

And India's specific position. A three-stage programme designed around thorium, of which India has among the world's largest reserves (Chapter 1.12), because it was denied uranium access for decades after 1974. The programme has progressed far more slowly than planned. The 2008 civil nuclear agreement restored international fuel access, and the domestic nuclear liability law has deterred foreign suppliers.

Solar

The cost story is the most remarkable in energy history. Photovoltaic module prices have fallen by around 99 percent since the 1970s and by around 90 percent in the last fifteen years.

Why. A learning curve — costs fall by a roughly constant percentage for each doubling of cumulative production — driven by manufacturing scale, mostly in China, and by incremental efficiency gains. Solar is now the cheapest source of new electricity in most of the world by levelised cost.

Its problems are real and are about the system rather than the panel.

Intermittency. It generates when the sun shines, which is not when demand peaks in most systems. The duck curve describes what happens on a grid with a lot of solar: net demand collapses in the middle of the day and then rises very steeply in the evening as solar output falls and household demand rises, requiring other generation to ramp fast.

Land. Utility-scale solar requires substantial area, which competes with agriculture — though rooftops, canal tops and degraded land are all available, and India has pursued all three.

Materials and manufacturing concentration (Chapter 12.10).

And end-of-life panel recycling, which is not yet at scale.

India's solar programme is among the world's largest, with very low auction prices achieved, and the constraints have been transmission capacity, land acquisition and payment reliability from state distribution companies rather than the technology.

Wind

Onshore wind is cheap and mature. Offshore wind is more expensive, more reliable, and has better capacity factors.

Problems: intermittency, though on a different pattern from solar, which makes the two complementary; local opposition on visual and noise grounds; bird and bat mortality, which is real and is far smaller than mortality from buildings, cats and vehicles; and grid connection.

India's wind resource is concentrated in Tamil Nadu, Gujarat, Karnataka and Maharashtra (Chapter 2.2).

Hydropower

The largest source of low-carbon electricity globally and the most mature.

Dispatchable, and pumped storage hydro is currently the dominant form of grid-scale energy storage.

And its costs are large and are borne locally. Displacement of populations — India's Narmada and Tehri projects displaced very large numbers, and the compensation and resettlement record is poor. Ecosystem disruption and blocked fish migration. Methane emissions from decomposing vegetation in tropical reservoirs, which can be substantial. Sediment trapping, which starves downstream deltas (Chapter 1.9).

And dam failure is the reason hydro's death toll is not lower. The Banqiao dam failure in China in 1975 killed an estimated 26,000 people directly and many more in the subsequent famine and epidemics.

The others

Geothermal, excellent where available — Iceland, Kenya, parts of the American west — and geographically limited, though enhanced geothermal systems could widen it.

Biomass. Carbon-neutral in principle if regrown and frequently not in practice, because the regrowth takes decades and the emissions are immediate. Traditional biomass burning for cooking is a major cause of indoor air pollution and a major cause of premature death, particularly among women and children in South Asia and Africa — which makes the transition to clean cooking fuel one of the highest-value health interventions available, and India's LPG connection programme is the largest such effort.

Tidal and wave, technically feasible and not yet economic at scale.

And efficiency, which is the cheapest source of all. The energy you do not use costs nothing and emits nothing. Building insulation, efficient appliances, industrial process improvement and reduced transmission losses are consistently the highest-return interventions and are consistently under-invested because nobody sells them.

Storage

The binding constraint on a high-renewable grid, and it deserves its own section because it is where the argument actually is.

Lithium-ion batteries have followed a cost curve similar to solar and are excellent for hours of storage — which handles the daily cycle.

They are not currently economic for seasonal storage — storing summer solar for winter — which is the harder problem in high-latitude systems and less of an issue in India.

Alternatives: pumped hydro, which is mature and geographically constrained; compressed air; thermal storage; and hydrogen, which is inefficient round-trip and is the leading candidate for long-duration and for industrial heat.

And the honest position. A grid with a very high share of variable renewables requires either enormous storage, a large geographic transmission network to smooth the variation, substantial dispatchable backup, or demand that can shift. Every real system uses a mixture, and the arguments about how much of each are technical questions with genuine disagreement.

The hard sectors

Electricity is the easy part. These are not.

Steel (Chapter 3.12), where the chemistry itself produces carbon dioxide and hydrogen reduction is the main alternative route.

Cement, where roughly half the emissions come from the chemical decomposition of limestone rather than from fuel, and where there is no drop-in substitute.

Aviation and shipping, where energy density requirements make batteries impractical for long distances.

And industrial heat at high temperatures.

Together these are a large share of emissions and their solutions are far less developed than power generation's.

Where this shows up in your life

Your electricity bill, and the fact that in most of India it is cross-subsidised — industrial and commercial users pay above cost so that agricultural and residential users pay below it, which is a political arrangement with real consequences for distribution company finances and therefore for grid investment.

And the framework worth taking from this chapter. Every energy source has costs; the question is never whether but which and borne by whom. Coal's costs are dispersed as air pollution deaths and future warming. Nuclear's are concentrated as rare accidents and long-lived waste. Hydro's fall on displaced communities. Solar's and wind's fall on land, materials and the system integration problem.

A source with no costs does not exist, and any advocate who cannot name their preferred option's disadvantages has not thought about it.

What the next page covers

Chapter 14.5 covers the transition — what has actually worked, what has not, why the Montreal Protocol succeeded and climate agreements have struggled, what carbon pricing does and where it has been implemented, what individual action does and does not achieve, and an honest assessment of where things stand.