Compression ignition engines remain essential for high energy-density applications; however, improving fuel energy utilization while limiting nitrogen oxide emissions remains challenging. This study develops a combustion-informed surrogate modeling and bounded multi-objective optimization framework for a fumigation-assisted compression ignition engine. Steady-state experimental data were integrated with physically derived combustion descriptors and modeled using gradient-boosting regression to predict brake thermal efficiency and regulated emissions. The surrogate model demonstrated reliable predictive capability for brake thermal efficiency and nitrogen oxide emissions, achieving coefficients of determination (R2) of 0.916 and 0.917, respectively, on the independent test dataset, thereby confirming good agreement between experimental observations and model predictions within the validated operating domain. Multi-objective optimization revealed a maximum efficiency improvement of 6–8% relative to baseline operation, accompanied by a 10–15% increase in nitrogen oxide emissions at peak efficiency conditions. A compromise operating region achieved a 4–5% efficiency improvement with less than 5% emission penalty. All optimized solutions were constrained within experimentally verified combustion stability limits to ensure physical feasibility. The proposed framework provides a systematic methodology for enhancing energy recovery under emission constraints and supports efficient combustion optimization strategies for advanced compression ignition engine systems.
M et al. (2026) studied this question.