Industrial heat electrification break-even

Levelised cost of heat from a resistive electric boiler1 against a natural-gas boiler, for process steam in Texas.

A chemical plant that wants to stop burning gas for process steam has one obvious substitute. Put an electric resistance heater where the burner was. No new chemistry, no new catalyst, nothing to invent. The whole question is arithmetic. What does the electricity have to cost before the substitution stops losing money?

The comparison is on delivered heat. Both routes are priced in dollars per million British thermal units that actually reach the process, not on fuel bought at the gate.

What the model finds

At an assumed gas price of $3.502 per million British thermal units and an assumed 6.5 cents per kilowatt hour for electricity, gas heat costs about $5.13 per million British thermal units and electric heat about $20.00. The break-even electricity price is about 1.53 cents per kilowatt hour, or 1.38 cents if only fuel is counted.

That threshold barely moves. Swinging capital cost and fixed operations across a range of plus or minus fifty per cent shifts the break-even by roughly 0.15 cents per kilowatt hour. The spark gap, meaning the ratio between the electricity price and the gas price, decides the outcome almost by itself. Arguments about equipment cost are arguments about the wrong variable.

Emissions give a second threshold. The electric boiler emits less than the gas boiler only when the grid carbon intensity3 is at or below 0.209 tonnes of carbon dioxide per megawatt hour. The ERCOT average is 0.333, from the most recent eGRID release, which reports 733.862 pounds per megawatt hour for the subregion. That is well above the threshold, so on average grid power electrifying this heat increases emissions: 99.6 kilograms per million Btu against the gas boiler’s 62.4.

Levelised cost of heat for the gas and electric boilers as stacked bars, split into fuel, levelised capital and fixed operations.

Where the money goes in each route. Fuel is almost the whole bar on both sides: $4.12 of the gas boiler’s $5.13, and $19.44 of the electric boiler’s $20.00. Capital and fixed operations come to $1.02 on the gas side and $0.56 on the electric, which is why the break-even below moves with the two fuel prices and almost nothing else.

Break-even electricity price as a function of the gas price. The line is the locus where the two routes cost the same; below it the electric boiler wins.

Break-even electricity price as a function of the gas price. The line is the locus where the two routes cost the same; below it the electric boiler wins.

Cost gap across the electricity price and gas price plane. The diagonal structure is the spark gap dominating both axes.

Cost gap across the electricity price and gas price plane. The diagonal structure is the spark gap dominating both axes.

Sensitivity of the break-even to every input. Fuel prices dominate; capital and operations barely move it.

Sensitivity of the break-even to every input. Fuel prices dominate; capital and operations barely move it.

Emissions of both routes against grid carbon intensity. The crossing point is the parity threshold.

Emissions of both routes against grid carbon intensity. The crossing point is the parity threshold.

Method

The levelised cost4 is built from an annualised capital charge, fixed and variable operations, and fuel, divided by annual delivered heat. Boiler efficiency is applied on the gas side, conversion and distribution losses on the electric side. The workbook carries the same arithmetic in live formulas, so the assumptions can be edited without touching code. The Python implementation is the reference for the other ports.

Limits

The prices in the shipped run are placeholders, and every figure above that depends on them is assumed rather than observed. They are set in the model, not read from data: heat/model.py holds them, and data/eia_price_template.csv beside it is the format for an EIA series, not a file the model loads. The boundary between retail all-in electricity and wholesale energy-only pricing has to be stated before any number is quoted, because it moves the answer by more than the capital cost does. Capital figures are scaling placeholders with a stated uncertainty band rather than vendor quotes. The emissions comparison uses average grid intensity. A marginal-emissions treatment is more defensible and is the next revision.

The code

Python reference implementation, Julia and Octave ports, and the formula-driven workbook. The archive holds the source only: no generated figures, no bulk data. Each model runs from its own README.md.

Download source All code

Sources

  1. Zuberi, Hasanbeigi & Morrow, Electrification of industrial process heat, Lawrence Berkeley National Laboratory, 2021.Electric boiler efficiency and capital cost ranges.
  2. U.S. Energy Information Administration, Electric Power Monthly Table 5.6.A and the Natural Gas Industrial Price series.The series to use for Texas industrial electricity and gas prices. The shipped run does not: its two prices are placeholders set in heat/model.py.
  3. U.S. EPA, eGRID2023 (revised 12 June 2025), ERCT subregion output emission rates; and EPA, Emission Factors for Greenhouse Gas Inventories, stationary combustion.Grid carbon intensity, 733.862 lb CO₂/MWh or 0.333 t/MWh, and the natural gas emission factor. The CO₂ rate is used rather than the CO₂e rate, because the gas factor it is compared against is also carbon dioxide only.
  4. NREL, Annual Technology Baseline 2024, financial assumptions.Discount rate, economic life and the levelisation method.