Waste Heat From Steel Furnaces Turned Into Electricity and Freshwater in Optimized New Design

Steel plants pour enormous amounts of heat into the atmosphere every time an electric arc furnace completes a tap-to-tap cycle, and a new modeling study argues that much of that discarded energy could be put to work twice over, generating both electricity and drinking-quality freshwater.

The starting point is the electric arc furnace itself. The study’s authors, Ali Nikravesh, Gholamreza Salehi, and Mohammad Ghotbi zadeh, modeled off-gas leaving the furnace at 600 degrees Celsius with a mass flow rate of 330 kilograms per second, composed mainly of carbon monoxide and carbon dioxide. Rather than feeding this erratic stream directly to turbines, they inserted a 150-cubic-meter steam accumulator between the heat recovery boiler and the downstream units, letting the vessel charge during hot phases and discharge steadily through the rest of the cycle.

Before optimizing anything, the team screened five candidate power cycles under identical heat-source and heat-sink conditions: a basic organic Rankine cycle, a recuperative version, a regenerative version with turbine bleed extraction, a combined regenerative-recuperative configuration, and a basic Kalina cycle running on an ammonia-water mixture. The organic cycles used R-123 as the working fluid. The regenerative-recuperative organic Rankine cycle, or RR-ORC, emerged as the strongest overall performer, delivering the highest net power output of the group at 2,031.95 kilowatts, the highest power-generation energy efficiency at 14.19 percent, and the highest power-generation exergy efficiency at 38.44 percent.

The economic picture reinforced that choice. The RR-ORC achieved the lowest levelized cost of electricity among the five configurations at 0.125 dollars per kilowatt-hour, the highest net present value at 1.307 million dollars, and the largest annual carbon credit gain, estimated at 220,290 dollars per year based on a grid emission factor of 0.364 kilograms of CO2 per kilowatt-hour and an assumed carbon trading price of 40 dollars per ton. No single configuration dominated every metric, however. The Kalina cycle recovered the largest fraction of available heat, 95.33 percent, and produced the cheapest freshwater at 3.388 dollars per cubic meter, while the recuperative ORC posted the shortest payback period at 9.991 years. The authors used these trade-offs to justify selecting the RR-ORC as the basis for the full optimization study.

That optimization was no small computational undertaking. The researchers coupled Aspen HYSYS process simulation with MATLAB through COM automation and ran the Non-dominated Sorting Genetic Algorithm II, better known as NSGA-II, over six decision variables: ORC turbine inlet pressure, degree of superheating, condenser saturation temperature, steam-accumulator charging pressure, MED feedwater flow rate, and top brine temperature. Six objectives were pursued simultaneously, maximizing net power output and the gained output ratio while minimizing total exergy destruction, the levelized cost of electricity, the levelized cost of water, and the levelized cost of storage. Ten independent algorithm runs with different random seeds were combined into a single Pareto front, and constraints on pinch-point temperature differences, turbine outlet vapor quality, subcritical operation, and pressure margins kept every candidate physically meaningful.

Because a Pareto front contains many equally valid trade-off solutions rather than one winner, the team applied the TOPSIS ranking method to identify a preferred compromise. The selected operating point used a turbine inlet pressure of 898.25 kilopascals, a modest superheating temperature of 7.88 degrees Celsius, a condenser temperature of 30.29 degrees Celsius, an accumulator charging pressure of 1,617.39 kilopascals, a MED feedwater flow of 28.06 kilograms per second, and a top brine temperature of 90.01 degrees Celsius. At this point the integrated system produced 2,077.10 kilowatts of net power with a gained output ratio of 7.94, meaning each kilogram of heating steam yields nearly eight kilograms of distillate. Compared with the unoptimized RR-ORC baseline, the optimized solution raised net power by 2.22 percent while cutting total exergy destruction by 2.29 percent, the levelized cost of electricity by 8.55 percent to 0.1143 dollars per kilowatt-hour, the levelized cost of water by 5.81 percent to 3.6924 dollars per cubic meter, and the levelized cost of storage by 14.61 percent to 0.007685 dollars per kilowatt-hour thermal.

Component-level exergy analysis pinpointed where the system bleeds the most useful work. The ORC condenser emerged as the single weakest link, with an exergy efficiency of just 12.67 percent and a relative cost difference of 112 percent, reflecting the irreversibility inherent in rejecting heat near ambient temperature. Within the desalination train, the MED condenser and the first effect were the dominant loss centers, posting exergy efficiencies of 47.9 percent and 60.17 percent respectively, while the remaining seven effects operated between 94.37 and 99.35 percent. The steam accumulator itself proved remarkably clean, achieving 98 percent exergy efficiency with a relative cost difference of only 2.36 percent, confirming that the buffering strategy adds little thermodynamic penalty of its own.

Sensitivity analysis revealed how sensitive the economics are to assumptions beyond the plant fence. Raising the interest rate from 9.6 to 14.4 percent pushed the levelized cost of electricity from 0.1023 to 0.1281 dollars per kilowatt-hour and the cost of water from 3.2845 to 4.1155 dollars per cubic meter. Hotter and faster off-gas streams, varied between 500 and 700 degrees Celsius and 280 to 380 kilograms per second, boosted power output and lowered electricity cost but simultaneously increased exergy destruction and raised the unit costs of freshwater and storage, underscoring that no single knob improves everything at once.

The dynamic simulation added a crucial reality check. Running the steam accumulator through the full 70-minute tap-to-tap temperature profile in Aspen HYSYS Dynamics, the researchers found that after an initial transient of roughly five minutes the vessel settled into a quasi-steady state, holding pressure near 1,315 kilopascals, temperature around 192.3 degrees Celsius, and liquid level close to 7.3 meters with only minor fluctuations. The authors caution that their dynamic scenario varied temperature only, holding mass flow constant at the nominal value, so simultaneous transients in both variables could produce larger deviations.

The study’s limitations are stated plainly. The comparative assessment and optimization rest on a steady-state, quasi-static framework representing average cyclic behavior, so the TOPSIS solution should be read as a nominal operating point rather than an instantaneous optimum. Seawater conditions were assumed constant, heat losses and pressure drops were neglected in the baseline model, and auxiliary facilities such as advanced gas cleaning and seawater pretreatment were excluded, meaning the reported economic indicators are comparative estimates rather than complete plant-level costs. Validation was performed at the subsystem level against published numerical benchmarks and industrial operating ranges, with model errors below 2 percent for the ORC configurations, below 4.1 percent for the Kalina cycle, and below 0.01 percent for the MED model, though the integrated system as a whole lacks independent experimental validation. Even with those caveats, the framework gives steel plants a concrete, quantified pathway for turning a notorious waste stream into two saleable products, and the authors point toward low-global-warming refrigerants, hybrid MED-RO desalination, and uncertainty-aware optimization as the natural next steps.

Subject of Research: Multi-objective optimization of a storage-assisted waste heat recovery system producing electricity and freshwater in electric arc furnace steel plants

Article Title: Multi-objective optimization of a storage-assisted waste heat recovery system for integrated and sustainable power and freshwater generation in steel plants

Article References: Nikravesh, A., Salehi, G., & Ghotbi zadeh, M. (2026). Multi-objective optimization of a storage-assisted waste heat recovery system for integrated and sustainable power and freshwater generation in steel plants. Results in Engineering, 32, Article 113545. https://doi.org/10.1016/j.rineng.2026.113545

Image Credits: AI Generated

DOI: 10.1016/j.rineng.2026.113545

Keywords: waste heat recovery, electric arc furnace, organic Rankine cycle, thermal energy storage, steam accumulator, multi-effect distillation, NSGA-II, TOPSIS, exergoeconomic analysis, steel industry, polygeneration, desalination