Multi-Scale Modeling and Environment-Safety-Economy Multi-Objective Optimization of Coke Dry Quenching System Based on Flow Fields Simulation and Deep Learning

Abstract Optimizing coke dry quenching operation is critical for energy saving, emissions reduction, and safety. However, the trade-off between economic benefit and CO2 emissions is constrained by localized combustible-gas accumulation, which is difficult to characterize using macroscopic indicators. To address this issue, a multiscale modeling and environment-safety-economy collaborative optimization framework is proposed. High-fidelity computational particle fluid dynamics (CPFD) results are used to construct a surrogate model predicting flow fields and macroscopic indicators. The surrogate model is further coupled with boiler-turbine mechanistic models, while maximum local CO and H2 volume fractions serve as safety constraints in multiobjective optimization of economic benefit and CO2 emissions. Results show that tightening the local H2 limit from 3% to 2% excludes high-risk operating conditions with excessive H2 accumulation, with the economic-benefit upper bound decreasing from 3.04 × 103 to 2.84 × 103 CNY/h. CPFD back-substitution gives maximum relative errors of 3.82%, confirming the framework reliability.