On a modeled cogeneration site, electrified steam raises total emissions and steam savings show up as electricity

CheAI Research Inc. | Research article | September 15, 2026

Keywords: electrification; electrode boiler; industrial heat pump; cogeneration; steam efficiency; Scope 2 emissions; grid losses

Abstract

1. Introduction

A steam efficiency measure is usually credited with the fuel its saved steam would have required [1]. That credit assumes the boilers respond to steam demand. At a cogeneration site they may not: when the turbines operate at their flow limits, the boilers must raise the steam the turbines take, and steam saved downstream is routed elsewhere in the network. An electrification measure is usually credited with the stack emissions it removes, which assumes the replacement electricity is carbon-free or omits it.

Optimal dispatch of this modeled site places every optimum on equipment limits, and at the cost optimum both turbines operate at maximum inlet flow. The present study examines the consequences for standard decarbonization measures. One configuration is changed at a time, the dispatch is re-optimized, and each result is compared with the documented baseline only, never with another scenario, so that each difference is attributable to a single configuration change. Process steam demand is held fixed in mass, not in delivered heat; where a scenario changes the steam's condition, the heat delivered to the process changes with it.

2. Methods

2.1 Baseline and comparison

The baseline is the optimized coupled schedule of the steam system and the chlor-alkali cell house: 3,331.28 USD per hour, importing 11.32 MW. A zero-change scenario solved through the same procedure reproduces it exactly in every field. Every result is reported as scenario minus baseline; all 20 difference records, central cases and sensitivity cases, refer to the same baseline.

2.2 Scenarios

The electrode boiler supplies the entire 70 t/h low-pressure process demand as saturated steam, taking 45.5 MW of electricity to deliver 45,030 kW of steam heat. The heat pump uses the cell house's rejected heat, unused in the baseline, with a source at 90 degrees C and a lift of 57.9 K; its compressor coefficient of performance of 3.08 is interpolated in lift from Danish Energy Agency technology data [2], reduced for steam generation, and lies below the Carnot bound of 7.27. Including auxiliary electricity, the system coefficient of performance is 2.90. The no-cogeneration case removes the turbines and lets steam down through valves; the backpressure-only case removes the condensing turbine; the minimum-tail case holds the condensing tail at its lower limit. Condensate return rises from 0.65 to 0.80, and blowdown heat is recovered into makeup water at 90 percent recovery. Each scenario re-solves the dispatch with the same endpoint treatment of the boiler split as the baseline. Steam properties follow IAPWS-IF97 [3], and grid emissions use the eGRID2023 CAMX factor [4], grossed up for grid losses. These grid emissions comprise Scope 2 and the grid-loss share, which the GHG Protocol assigns to Scope 3 [5].

2.3 Physical checks

Every scenario's solution satisfies the same mass, energy and second-law balance checks as the base network; the largest residual as a fraction of its tolerance is 0.070. Heat-pump coefficients of performance are checked against the Carnot bound and a lift limit; no operating point violates either.

3. Results

Table 1 and Figure 1 give each scenario's change against the baseline per hour of operation.

Table 1. Change from baseline for each scenario, per hour.

3.1 Electrified steam raises total emissions on this grid

The electrode boiler removes 125.2 GJ/h of boiler fuel and 6,297 kg CO2 per hour from the stack, and raises grid import by 41.33 MW, which adds 8,406 kg CO2e per hour at the delivered CAMX factor. The net change is an increase of 2,110 kg per hour. Counted at the stack alone, electrification would appear as a large reduction; counted gate to gate, it is an increase. With the re-optimized dispatch held fixed, the net change crosses zero at a delivered grid intensity of 152 kg CO2e per MWh, against the delivered CAMX average of 203 (Figure 2).

The break-even depends on the emission factor used. An annual average describes the grid's entire output, whereas the electricity an electrode boiler adds is supplied at the margin, whose intensity varies by hour [6].

3.2 Removing cogeneration raises cost but barely changes emissions

Without the turbines, the site purchases all its electricity and lets steam down through valves. Cost rises by 771.56 USD per hour. Scope 1 falls by 2,839 kg per hour as boiler fuel decreases, and grid emissions rise by 2,781 kg per hour, a net reduction of only 58 kg per hour, 0.2 percent of the baseline. Removing cogeneration removes both the backpressure turbine, whose power is generated from steam the processes require anyway, and the carbon-intensive condensing turbine; together they carry about as much carbon per megawatt-hour as the grid power that replaces them. On this grid, removing cogeneration raises cost but barely changes combined inventory emissions; on a more carbon-intensive grid, cogeneration would also reduce emissions.

3.3 Steam savings appear as electricity, not fuel

Improved condensate return saves 28.75 USD per hour and blowdown recovery 48.43 USD per hour. Both change boiler fuel by exactly zero. The boilers raise the steam that the turbines, at their flow limits, take; a measure that reduces makeup heating or deaerator steam frees steam inside the network rather than at the boiler, and the optimum routes it through the condensing tail. Import falls by 0.38 and 0.65 MW respectively. The savings are real, but they are electricity savings.

The heat pump shows the effect most clearly. It upgrades the cell house's waste heat into 8.56 t/h of low-pressure steam, delivering 5,504 kW of heat for 1.90 MW of electricity. Because boiler output does not change, this steam displaces letdown flow and passes to the condensing tail, which rises from 14.2 to 21.9 t/h and generates 1.44 MW more. That is less than the heat pump consumes, so import rises by 0.45 MW, cost by 34.01 USD per hour and grid emissions by 92 kg per hour.

3.4 Comparison with the DOE bandwidth estimate

The DOE Bandwidth Study gives a practical-minimum opportunity of 2.7 percent for crosscutting utilities in chemical manufacturing [7]. That is a sector-wide crosscutting estimate, not a steam-only limit or a target for this plant, and the scenarios here are not comparable to it as a quantity. The single illustrative scale-up, pre-registered before any result existed, multiplies the condensate-return scenario's fractional change in boiler gas use by the sector's 1,440 trillion Btu per year of boiler natural gas [8]. Because that fractional change is exactly zero, the illustrative result is 0.0 trillion Btu per year. The result is degenerate and is reported as such.

4. Discussion

Two structural properties explain every result. First, gate-to-gate accounting changes the sign of electrification's effect: on a grid at the CAMX average, moving steam from gas to electricity shifts emissions from the stack to the power sector and increases them. Second, when turbine limits rather than steam demand set boiler output, steam-side measures act as electricity measures. Their value then depends on the price and carbon intensity of grid power rather than gas, and a fuel-based estimate of their savings is incorrect in kind, not only in magnitude.

Neither property is general. A site whose turbines are not at their limits would convert saved steam into saved fuel, and a substantially cleaner grid would favor the electrode boiler. Hourly marginal accounting would not change the conclusion here, since California's 2023 marginal emission rate exceeded the annual average in every hour. The contribution is the mechanism, and a gate-to-gate model that makes it visible.

5. Limitations

• Results apply to one modeled configuration and its documented baseline, not to a real site, the US chemical sector or national savings potential.
• The break-even intensity is computed with an annual-average factor and does not apply to any particular hour.
• The heat-pump coefficient of performance takes the source at the cell temperature with no approach, an optimistic bound.
• Process steam demand is fixed in mass. The electrode boiler delivers saturated steam where the baseline low-pressure header carries strongly superheated steam, so less heat reaches the process in that scenario; a comparison at fixed useful heat would require additional electrode-boiler output.
• Unmodeled variants, such as electric superheating, were not evaluated.

6. Conclusions

On the modeled site and the CAMX grid, an electrode boiler raises total emissions by 2,110 kg per hour and breaks even only below 152 kg CO2e per MWh delivered. Because turbine limits fix boiler output, condensate, blowdown and heat-pump measures save electricity rather than fuel. Steam-efficiency and electrification assessments at cogeneration sites should therefore establish what sets boiler output and count emissions gate to gate.

References

1. U.S. Department of Energy, Advanced Manufacturing Office. Steam Tip Sheets 8, 10 and 12: Return Condensate to the Boiler (DOE/GO-102012-3410); Recover Heat from Boiler Blowdown (DOE/GO-102012-3408); Flash High-Pressure Condensate to Regenerate Low-Pressure Steam (DOE/GO-102012-3400). Washington, DC, 2012.
2. Danish Energy Agency. Technology Data for Industrial Process Heat, data sheet, 2025 central values. Copenhagen, 2026.
3. International Association for the Properties of Water and Steam. Revised Release on the IAPWS Industrial Formulation 1997 for the Thermodynamic Properties of Water and Steam, IAPWS R7-97(2012). 2012.
4. U.S. Environmental Protection Agency. Emissions and Generation Resource Integrated Database, eGRID2023 Revision 2. Washington, DC.
5. World Resources Institute and World Business Council for Sustainable Development. GHG Protocol Scope 2 Guidance. Washington, DC, 2015.
6. Siler-Evans K, Azevedo IL, Morgan MG. Marginal emissions factors for the U.S. electricity system. Environmental Science and Technology, 46(9):4742-4748, 2012.
7. U.S. Department of Energy. Bandwidth Study on Energy Use and Potential Energy Saving Opportunities in U.S. Chemical Manufacturing. Washington, DC, 2015.
8. U.S. Energy Information Administration. Manufacturing Energy Consumption Survey 2022: Data Tables. Washington, DC.

Related studies

• Optimal dispatch of the modeled site: why the turbines operate at their limits
• Inventory and avoided emissions: annual and hourly grid factors compared
• Verification of DOE steam tip sheets: the fuel-based savings arithmetic
• Chlor-alkali and steam system coupling: the heat source for the heat pump