Cogeneration system optimization in refineries confronts the challenge of simultaneously integrating design parameter selection and topological configuration. The literature typically addresses these aspects separately: parametric optimization with fixed topology or configuration optimization for specific nominal conditions. This work develops a comprehensive methodology integrating exhaustive parametric exploration with superstructure-based optimization through mixed-integer nonlinear programming (MINLP), applied to the Miguel Hidalgo refinery in Tula, Mexico. The systematic procedure generates superstructures considering all viable expansion and tempering routes under steam quality restrictions (x≥0.88), evaluating 84–105 combinations of generation pressure (PHRSG=70–140 bar) and superheater outlet temperature (Ts4=500–560 °C). The analysis reveals three topologically distinct configurations identified as generating maximum power under different operating conditions and characterizes how transitions between high-performing configurations occur at discrete thermodynamic thresholds that correlate with constraint activation contradicting the conventional assumption of continuous parameter-configuration relationships. Multi-criteria evaluation positions Configuration 1 as the recommended design, generating 25% increase in electric generation, 11% improvement in utilization factor (UF: 0.640→0.710) and 20% reduction in specific fuel consumption (SFC: 0.259→0.207 kg/kWh). The methodology is directly generalizable to other refineries through universal thermodynamic principles, with a systematic five-step procedure applicable to any multi-pressure steam demand profile. The characterization of discrete transition phenomena and the associated methodology for their thermodynamic explanation challenges the conventional assumption of continuous parameter–configuration relationships in optimization approaches, with immediate implications for the design of flexible cogeneration systems in refineries worldwide.
Pereyra et al. (Mon,) studied this question.