2004 Control and diagnosis oriented modelling of the compression ignition engine

Olivier Grondin
Defended on December 17, 2004.

Olivier Grondin was born in 1977. He received a Bachelor degree in 1995. From 1995 to 1997, he was studying power electronics. He obtained upper technician diploma (BTS) in 1997. Subsequently, he was involved in electronics and signal processing until 2000. As the result of this work, he received a Master’s degree in Electronics from the University of Poitiers. In September 2000, he was involved in the DEA T3IA (Traitement des Informations : Informatique, Images et Automatique) with automatic control speciality. He was done a six month training period in the LAII (Automatic Control Laboratory of Poitiers) on active harmonics filtering (for power plants).

In September 2001, he was selected for a Ph’D thesis on the following subject : Control and diagnosis for Internal Combustion Engines. His thesis was defended on December 17, 2004. He spent a post-doctoral position at Institut Français du Pétrole.

PDF - 6.9 Mb

Abstract This thesis has described an investigation into the modelling of compression ignition engine for control and diagnosis purpose. The Diesel engine is the most efficient and clean internal combustion engine due to modern electromechanical actuators. However, pollutant emission regulations are much more stricter, thus, these complex systems need sophisticated and efficient control algorithms to reach very low emission levels. For this task, engine models are required at each step of the control system development : control laws synthesis, simulation and validation. The system under study is a six cylinder direct injection Diesel engine fitted with a turbocharger. The model of this system is based on physical laws for some parts of the engine such as cylinders, manifolds, turbocharger and crank-slider system. In order to reduce computing time heat transfer and heat release during combustion were modeled using simple empirical correlations. The resulting model has been implemented in the Matlab-Simulink environment and it can predict variables of interest for control purpose with one degree crank angle resolution. It has been tested numerically and compared with an industrial engine simulation code with good results. Moreover, model output variables are in good agreement with experimental data recorded on a heavy-duty research engine. The engine model has been embedded on a board providing enough computing performances to perform real-time simulations, this will be helpful for “hardware-in-the-loop” simulations. Another part of this study is dedicated to the combustion process modelling using a non linear phenomenological model : the NARMAX model. The goal is to predict the in-cylinder pressure evolution using other measurements available on the engine. The NARMAX model parameters have been identified using input-output data carried out from the experimental engine. Such model is well suited for real-time applications compare to numerically cost effective physical models.

- References :

O. Grondin, R. Stobart, H. Chafouk & J. Maquet, Modelling the compression ignition engine for control : Review and future trends, SAE Technical Paper 2004-01-423.


PDF · 6.9 Mb
12421 - 24/03/17

ATOMOSYD © 2007-2017 |  Follow-up of the site's activity  |  SPIP  |  scoty  |  MàJ . 06/02/2017