Modelling aspects of a CNG injection system to predict its behavior under steady state conditions and throughout driving cycle simulations

Alternative fuels such as compressed natural gas (CNG) have lately gained an increasing interest due to the even more stringent emission regulations and environmental policies. Considerable R&D activity has been carried out in order to enhance the mixture formation and to improve the after treatment efficiency. The achievement of these goals cannot disregard the importance of the injection system and of its component (common rail, injectors and pressure reducing valve) on the air-to-fuel mixing. The amount of injected fuel should in fact be properly targeted by the ECU basing on the estimation of the induced air and accounting for the embedded closed-loop strategies. Still, these latter are normally derived from engine-base routines and totally ignore the injection system dynamics. The present paper aims at developing a thorough model of the injection system with specific attention to the one component which significantly contributes to fully defining its dynamic response, i.e. the pressure reducing valve. The pressure reducer is made up of various elements that retain diverse weights on the valve behavior and should consequently be differently addressed to. The research has been focused on defining the set of parameters to be precisely reproduced in the 0D-1D simulation so as to match the injection system experimental behavior. A refined model of the pressure reducer has hence been proposed and the model has been calibrated, tested and run under various operating conditions so as to assess for the set-up validity. Comparisons have been carried out on steady state points as well as through out a vehicle driving cycle and the model capability to properly reproduce the real system characteristic has been investigated into. The proposed valve model has proved to consistently replicate the injection system response for different speed and load conditions. A few methodological indications concerning modeling aspects of a pressure regulator can be drawn from the present study. Finally, the model has been run in a predictive mode so as to inquiry into the response of the system to fast transient operations, both in terms of speed and load. The model outputs have highlighted mismatches between the ECU target mass and the actually injected one and have hinted at the need for dedicated and refined control strategies capable of preventing anomalies in the mixture formation and hence in the engine functioning. (C) 2015 Elsevier B.V. All rights reserved.