A model reduction case study: automotive engine air path

Low complexity plant models are essential for model based control design. Often a detailed high order model is available and simplification to a low order approximative model is needed. This paper presents a case study of two model reduction methodologies applied on the automotive engine air path. The first methodology is based on balanced truncation of models obtained by linearization around equilibria and trajectories. Under appropriate assumptions, this technique yields strict bounds on the approximation error. The second is a heuristic methodology, based on intuition commonly used when modeling engine dynamics. Although it is successfully used in practice, the approximation error is seldom known. The two methodologies are used to derive simple models for the required fuel charge in an SI engine, given engine speed and throttle positions. Performance, complexity and similarities of the two resulting low order models are compared