Multi-Objective Optimization and Reduced-Order Modeling for Aftertreatment System Performance Prediction

<div class="section abstract"><div class="htmlview paragraph">As emission regulations grow increasingly stringent, aftertreatment system designs are becoming more complex, requiring robust performance across the full range of Engine Operating Points (EOPs). Traditionally, aftertreatment development has relied on Computational Fluid Dynamics (CFD) simulations conducted at a limited set of representative points, focusing primarily on single performance objectives such as minimizing back pressure, enhancing ammonia uniformity, or reducing Diesel Exhaust Fluid (DEF) deposits. However, these objectives are inherently interdependent, and optimizing for a single parameter often negatively impacts others, leading to suboptimal system performance across the full engine map.</div><div class="htmlview paragraph">To address these challenges, this paper presents a multi-objective optimization framework combined with a reduced-order modeling approach to predict aftertreatment system behavior across the full engine operating range. The methodology captures the interactions among various performance parameters, providing a more holistic basis for system evaluation and design decisions. By integrating multi-objective optimization with predictive modeling, the approach improves analysis efficiency and supports more informed design trade-offs compared to traditional single-objective, limited-point simulations.</div></div>