Investigation of an efficient model-based fuel economy optimization methodology for diesel engines focusing on real world driving emission cycles

Together with stringent emission standards, meeting fuel consumption, and performance targets have made engine calibration process complicated, lengthy, and costly. For this reason, it is crucial to develop an efficient methodology for the optimization of a diesel engine calibration. In this study, a novel automated fine tuning methodology using genetic algorithms and data driven models is proposed. Fuel consumption of a light commercial vehicle with 2.0-L diesel engine was optimized for a predefined Real Driving Emissions (RDE) cycle. During the optimization, tail pipe emission limits were maintained by taking the aftertreatment system efficiency into account. To this end, a representative engine control unit, internal combustion engine, and Selective Catalyst Reduction (SCR) system models were built by using empirical data and neural networks. Moreover, the effect of choosing different tailored optimization points for a predefined route was investigated by comparison of frequency and fuel consumption based weighting. The cost function, related to the optimization problem, was defined based on fuel consumption, performance, and mechanical limits of the engine, and in accordance with tailpipe emission limits defined by Euro 6d-Temp regulations. Compared to previous model-based calibration studies, tailpipe emissions, and regulation limits have been incorporated in this study in order to obtain more realistic calibrations. After testing different genetic algorithm configurations, the fuel efficiency over the predefined route was improved by 3.07% without violating any of the limits while the smoothness of the calibration maps was also preserved. This route specific calibration methodology is believed to have the potential to increase fuel economy in real world applications such as daily commuting routes or repetitive driving patterns for instance, operation vehicle use cases in factories, airports, and construction areas. As for further development, this can be achieved by running simulations and the optimization on cloud and updating the calibration remotely based on the selected route before the trip.