Analytical Target Cascading Framework for Diesel Engine Calibration Optimisation

Abstract

This paper presents the development and implementation of an Analytical Target Cascading (ATC) Multi-disciplinary Design Optimisation (MDO) framework for the steady state engine calibration optimisation problem. The case is made that the ATC offers a convenient framework for the engine calibration optimisation problem based on steady state engine test data collected at specified engine speed / load points, which is naturally structured on 2 hierarchical levels: the ‘Global’ level, associated with performance over a drive cycle, and ‘Local’ level, relating to engine operation at each speed / load point. The case study of a diesel engine was considered to study the application of the ATC framework to a calibration optimisation problem. The paper describes the analysis and mathematical formulation of the diesel engine calibration optimisation as an ATC framework, and its Matlab implementation with gradient based and evolutionary optimisation algorithms. The results and performance of the ATC are discussed comparatively with the benchmark steady state solution for the engineering calibration of the diesel engine. The main conclusion from this research is that ATC optimisation framework offers an effective approach for engine calibration, with a potential to deliver significant fuel economy benefits. Further advantages of the ATC framework is that it is flexible and scalable to the complexity of the calibration problem, and enables calibrator preference to be incorporated a priori in the optimisation problem formulation, delivering important time saving for the overall calibration development process.