Loading Events

Dissertation Defense

Limitations and Improvements in the Idle Speed Control of a Direct Injection Spark Ignition Engine

Joseph James Scillieri


One of the aims of automotive powertrain research is to balance the achieve often-conflicting goals of performance, emissions control, and fuel economy. One outcome of this research is the direct injection spark ignition (DISI) stratified charge engine. This engine can run in multiple operating modes, including a lean burn, stratified charge mode with air-fuel ratio of 40-50:1. Running in stratified mode results in improved fuel economy and reduced carbon dioxide emissions. The stratified charge mode is often employed during low speed and load conditions, such as during engine idle.

This advanced technology engine has introduced a number of challenging control problems. However, the literature does not contain a control-oriented model, complete with parameter values, for an engine of this type. We construct a generic, control-oriented model based on existing, published conventional engine models and a simple heuristic argument. Results using the model are compared to existing results in the literature.

The idle speed control problem is considered next. This problem is cast as a two-input-two-output control problem, and a control topology from the literature that partially decouples the coupled TITO plant is examined. Sensitivity and robustness issues of a decoupling control design are introduced. We develop a baseline idle speed controller via decoupled loop closure. However, significant delays inhibit our ability to improve the transient response of the engine speed and in-cylinder air-fuel ratio via feedback alone. A multivariable scheme employing reference feedforward is proposed, and several potential topologies are presented. A reference feedforward algorithm is derived, and nonlinear simulation results are shown in which the speed and air-fuel ratio responses are improved considerably. However, the effectiveness of reference feedforward in improving the transient response can be limited by delays.

Sponsored by

James S. Freudenberg and Jessy Grizzle