Reduced-Order Long-Horizon Predictive Thermal Management for Diesel Engine Aftertreatment Systems

Abstract

Model predictive thermal management of a heavy duty diesel engine aftertreatment system (ATS) requires optimization over a long horizon due to slow thermal dynamics of the ceramic-based catalysts. A model-based optimization architecture is presented to decouple the diesel engine air path and torque with fast dynamics from the ATS slow thermal dynamics and simplify solving the thermal management optimal control problem (OCP). The key idea presented is to estimate the air path variables with static models and control the engine torque within an internal control loop. Then, an ATS predictive thermal management system is proposed as a high-level controller and a reduced order OCP is formulated with only ATS temperatures as states and two control variables, namely the engine intake manifold pressure and fuel injection timing. Simulation results over the FTP drive cycle indicate from 3 to 4% increase in the cycle-averaged diesel oxidation catalyst (DOC) and selective catalytic reduction (SCR) system efficiency, depending on the catalyst location and length of the controller's prediction horizon, along with 2-3% brake specific fuel consumption reduction over the test cycle. Results also show long-term predictions are required to avoid temperature drop after warm-up which happens during low exhaust temperature operations such as a long engine idle.