The Substitution Atlas
Fuel switching: who can defect from what
With single-fuel cushions gone, the fossil system's remaining slack is the ability to swap molecules. This atlas maps the degrees of freedom: where energy is used, how much of each use can switch to another source, on what timescale. The matrix below is the measured base; the switchability layer is curated from literature with explicit timescale classes and cited constraints.
How to read this page: ● measured sourced data · ◐ inferred analyst reading, basis linked · ○ projected anchored to a real starting point. Bracketed citations link to the sources at the foot of the page.
EU-27 energy consumption by sector and fuel (2024)
| Sector | Total | Oil | Gas | Coal | RE | Elec | Heat |
|---|---|---|---|---|---|---|---|
| Total final consumption | 878 | 328 | 172 | 12 | 109 | 206 | 40 |
| Industry (total) | 216 | 23 | 68 | 7 | 25 | 72 | 12 |
| Transport (total) | 284 | 257 | 3 | 0 | 17 | 7 | — |
| Road transport | 267 | 246 | 2 | — | 17 | 2 | — |
| Rail transport | 5 | 1 | — | 0 | 0 | 4 | — |
| Domestic aviation | 6 | 6 | — | — | 0 | — | — |
| Households | 228 | 24 | 67 | 4 | 52 | 61 | 19 |
| Commercial and public services | 119 | 7 | 30 | 0 | 10 | 62 | 9 |
| Agriculture | 26 | 15 | 3 | 0 | 3 | 4 | 0 |
| Power generation (transformation input) | 493 | 12 | 82 | 66 | 156 | 5 | 1 |
| Non-energy use (feedstock) | 78 | 66 | 11 | 1 | 0 | — | — |
Values in Mtoe. Source: Eurostat (nrg_bal_c), 2024.
Switchability by use
How much of each use can switch fuels, and how fast. Every share is a cited range, never computed pseudo-precision. Timescale classes: days (operational), weeks (logistics), months (minor capital), years (major capital). Unmapped cells render explicitly.
Power generation
493 MtoeDual-fuel gas turbines and steam plants can switch to gasoil or coal where both boilers exist.
Industry (process heat)
216 MtoeDual-fuel industrial boilers can switch between gas and gasoil.
Road transport
257 MtoeFleet turnover is the only mechanism.
Aviation
Near-zero short-term fuel switching.
Petrochemical feedstock
Steam crackers can switch between naphtha, LPG, and ethane feeds within weeks (feed flexibility is a design feature).
Residential heating
228 MtoeGas boiler replacement with heat pumps is years-timescale (capital, installation, grid capacity).
Shipping bunkers
LNG-fuelled fleet is ~1% of global fleet (by number, higher by capacity).
US refinery configuration classes
Each refinery classified by its most complex downstream unit. Configuration determines what crude quality the refinery can process: cokers handle heavy-sour (Maya, Arab Heavy), crackers convert middle distillates, hydroskimmers run only light-sweet.
coking
12,604 kb/d
50 facilities (69%)
cracking
3,988 kb/d
36 facilities (22%)
hydroskimming
742 kb/d
25 facilities (4%)
hydrotreating
459 kb/d
9 facilities (3%)
hydrocracking
375 kb/d
4 facilities (2%)
US only (EIA Form 820, Jan 2026).124 facilities, 18,168 kb/d total. Coverage: 100% US; rest-of-world classification staged (regional aggregates + per-facility curation in progress).
Switching price signals
The physical capacity to switch is one dimension; the economic trigger is the other. These are the current price levels that drive switching decisions.
Henry Hub (US gas)
$2.86/MMBtu
2026-08 | Below $4 = gas cheap, slack ample
EUA (carbon cost)
€82.61/tCO₂
2026-08-24 | Drives coal-gas switching parity in EU power
Related
Methodology
The sector × fuel matrix is measured from Eurostat complete energy balances (nrg_bal_c, EU-27, annual, CC BY 4.0). The switchability layer is curated: every share is a cited range from IEA, academic literature, or industry sources, never a model output presented as data. Timescale classes distinguish operational switching (days) from capital-intensive transitions (years). Cells without a credible estimate render as "unmapped".
The refinery configuration classification uses EIA Form 820 per-facility data (US, annual, public domain). Each facility is classified by its most complex downstream unit. Rest-of-world classification uses regional aggregates from national statistics and is staged in descending capacity order.
Three epistemic tiers: measured (Eurostat volumes, EIA capacities), inferred (switchability shares as cited ranges), and the flexibility lens verdict on the Tension Board (openly qualitative, computed from HH price and system-wide slack assessment).
Sources and method (5)
- [1] Eurostat energy balances (nrg_bal_c)
- [2] EIA Refinery Capacity Report (Form 820)
- [3] IEA World Energy Outlook 2024
- [4] EIA Henry Hub (STEO)
- [5] EUA front-month (investing.com)
Energy balances: Eurostat nrg_bal_c (CC BY 4.0). Refinery capacity: EIA Form 820 (public domain). Switchability: curated from IEA WEO 2024, industry sources. Gas price: EIA STEO. Carbon: investing.com (EUA front-month).