Search Results

The introduction of transient test cycles and the focus on real world driving emissions has increased the importance of ensuring the NOx and soot emissions are controlled during transient manoeuvres. At the same time, there is a drive to reduce the number of calibration variables used by engine control strategies to reduce development effort and costs. In this paper, a control orientated combustion model, [1], and model predictive control strategy, [2], that were developed in simulation and reported in earlier papers, are applied to a Diesel engine and demonstrated in a test vehicle. The paper describes how the control approach developed in simulation was implemented in embedded hardware, using an FPGA to accelerate the emissions calculations. The development of the predictive controller includes the application of a simplified optimisation algorithm to enable a real-time calculation in the test vehicle.

Setting up engines to meet emissions limits often involves extensive steady-state calibration activities combined with ad-hoc strategies to compensate for transient operation. As engines become more complex and acceptable emissions levels ever lower, this task is becoming increasingly time consuming and expensive. The inclusion of models in the engine control units offers a way to reduce some of this calibration effort. Model-based control is an active area of research with advanced approaches now being proposed. One example is the use of real-time models to regulate the burn angle during transient manœuvres. This paper describes the application of a control-orientated combustion model to control directly emissions during transients. The model is used to optimize and constrain the NOx emissions directly, rather than controlling an inferred variable such as the burn angle. This has the benefit that calibration engineers will be able to set the emissions trade-off directly.