Development of a Simulation based Powertrain Design Framework for Evaluation of Transient Soot Emissions from Diesel Engine Vehicles.

Abstract

This dissertation presents the development of a modeling and simulation framework for diesel engine vehicles to enable soot emissions as a constraint in powertrain design and control. To this end, numerically efficient models for predicting temporallyresolved transient soot emissions are identified in the form of a third-order dual-input single-output (DISO) Volterra series from transient soot data recorded by integrating real-time (RT) vehicle level models in Engine-in-the-loop (EIL) experiments. It is shown that the prediction accuracy of transient soot significantly improves over the steady-state maps, while the model remains computationally efficient for systemslevel work.

The evaluation of powertrain design also requires a systematic procedure for dealing with the issue that drivers potentially adapt their driving styles to a given design. In order to evaluate the implications of different powertrain design changes on transient soot production it is essential to compare these design changes on a consistent basis. This problem is explored in the context of longitudinal motion of a vehicle following a standard drive-cycle repeatedly. This dissertation develops a proportional-derivative (PD) type iterative learning based algorithm to synthesize driver actuator inputs that seek to minimize soot emissions using the Volterra series based transient soot models. The solution is compared to the one obtained using linear programming. Results show that about 19% reduction in total soot can be achieved for the powertrain design considered in about 40 iterations.

The two contributions of this dissertation: development of computationally efficient system level transient soot models and the synthesis of driver inputs via iterative learning for reducing soot, both contribute to improving the art of modeling and simulation for diesel powertrain design and control.