Emission control concepts for connected Diesel powertrains
- Emissionsregelungskonzepte für vernetzte Antriebsstränge mit Dieselmotor
Vagnoni, Giovanni; Pischinger, Stefan (Thesis advisor); Abel, Dirk (Thesis advisor)
Dissertation / PhD Thesis
Dissertation, Rheinisch-Westfälische Technische Hochschule Aachen, 2021
The increasing connectivity of future vehicles allows the prediction of the powertrain operational proﬁles. This technology can potentially improve the control of the engine and its exhaust gas aftertreatment systems. The study describes the development of rule- and optimization-based algorithms, which use the a-priori knowledge of upcoming driving events to reduce especially nitrogen oxides (NOx) and particulate (soot) emissions. In the ﬁrst part of the work, the boosting, the Lean NOx Trap (LNT) and the Diesel Particulate Filter (DPF) systems of a diesel powertrain are investigated as relevant subsystems for a typical passenger car application. Reference control strategies, based on state-of-the-art Engine Control Unit (ECU) algorithms and suitable predictive control logics, are compared for the three subsystems in a Model in the Loop (MiL) simulation environment. The simulation driving cycles are based on Worldwide harmonized Light duty Test Cycle (WLTC) and Real Driving Emissions (RDE) proﬁles. WLTC simulation results show an improvement potential for engine-out soot and NOx emissions of up to 5.5 % and 4.9 % respectively for the air path case. Additionally, the developed rule-based algorithm allows the adjustment of the NOx-soot trade-oﬀ, while keeping the fuel consumption constant. A reduction of the average fuel consumption in RDE of up to 1 % for the LNT case is achieved, thanks to the avoidance of aborted regeneration events. Similarly, also the DPF regeneration process is improved, sparing up to 5.5 % fuel in a representative real driving mission. In the second part of the work, a concept for an Integrated Engine and Exhaust Aftertreatment System Supervisory Controller is proposed for a conventional long-haul truck. It relies on a Nonlinear Model Predictive Control (NMPC), whose simpliﬁed Optimal Control Problem (OCP) formulation allows its real-time application and reduces its calibration eﬀort. The concept is benchmarked in the simulation environment against Dynamic Programming (DP) techniques and ﬁnally validated at the engine test-bench. Measurement results show the effectiveness of the developed controller in minimizing the powertrain operational costs, while complying with the emission constraints at the tailpipe. The work concludes with brief recommendations for future research directions such as the introduction of a prediction module for the estimation of the vehicle operational proﬁle in the prediction horizon and the extension of the developed algorithms to electriﬁed diesel powertrains.