Design and Experimental Validation of a Model-Based Injection Pressure Controller in a Common Rail System for GDI Engine Alessandro di Gaeta, Giovanni Fiengo, Angelo Palladino and Veniero Giglio Abstract— Progressive reductions in vehicle emission re- quirements have forced the automotive industry to invest in research and development of alternative control strategies. All control features and resources are permanently active in an electronic control unit, ensuring the best performance in terms of pollutant emissions and power density, as well as driveability and diagnostics. A way to attain these goals is the adoption of Gasoline Direct Injection (GDI) engine technology. In order to assist the engine management system design, through a better performance of GDI engine and the Common Rail (CR) system, in this work an injection pressure regulation to stabilize the fuel pressure in the CR fuel line is proposed and validated via experiments. The resulting control strategy is composed by a feedback integral action and a static model-based feed-forward action whose gains are scheduled as function of fundamental plant parameters. The tuning of the closed loop performance is then supported by an analysis of the phase margin and the sensitivity function. Preliminary experimental results confirm the effectiveness of the control algorithm in regulating the mean value rail pressure independently from engine working conditions, i.e. engine speed and time of injection, with limited design effort. I. INTRODUCTION The CR injection system technology has been originally introduced for diesel engines in order to achieve both the reduction of pollutant emissions enforced by international regulations and the improvement of performance required by the customers. The key device of this system is the common rail, i.e. a steel manifold where the fuel is kept at high pressure. The electronically controlled high pressure fuel injection system holds an important role concerning both the emission control strategy and the improvement of internal combustion engine performance [1], [2]. High pressure injection allows to finely atomize the fuel spray and to promote fuel and air mixing, resulting in significant combustion improvements [3], [4]. Hence the control of the injection pressure plays a fundamental role to achieve a better and better engine performance by respecting the current stringent emission regulations. Recently, this technology has been extended to gasoline engines [5], [6] and significant improvements have been reached adopting more sophisticated and precise control methodologies. As an example, in [7] a controller is designed applying the Quantitative Feedback Theory (QFT) to the This work was supported by the Italian Ministry for University and Research in the framework of the FIRB projects. A. di Gaeta and V. Giglio are with Istituto Motori, National Research Council, 80125 Napoli, Italy. E-mail: {a.digaeta, v.giglio}@im.cnr.it. G. Fiengo and A. Palladino are with the Dipartimento di Ingegne- ria, University of Sannio, 82100 Benevento, Italy. E-mail: {gifiengo, an- gelo.palladino}@unisannio.it. closed-loop system. It results in a controller robust to model uncertainties and external disturbances, having moreover a quantitative measure of the robustness achieved. Simulations and experiments on a one-cylinder diesel-dual-fuel engine have shown interesting performance of the proposed con- troller, compared to the traditional PID. In [8] an hybrid model of the Magneti Marelli Powertrain common-rail fuel- injection system for four-cylinders multijet engine has been presented. The hybrid controller is then compared with a classical regulator designed via mean value based approach. Even though an increasing number of technical papers proposing advanced feedback control loop are available, the control of high pressure in commercial products is still based on open loop strategy based on open loop strategy using tables function of pressure. Using the open loop control, injection pressure value is selected directly according to the engine operating conditions. Obviously this approach requires a non negligible time and material resources to identify the pressure mapping, and the resulting maps need to be updated in order to fulfill future emission reduction and fuel economy requirements. On the other hand, when reliable and simple enough models of the pressure rail are available, costs and resources can be reduced by using model-based control strategies. In [9] authors proposed a control oriented model of the rail pressure in the case of a 2 liters GDI spark ignition engine. The model is capable to predict the mean value rail pressure as a good compromise between accuracy and model complexity, resulting particulary amenable for the design of efficient control algorithms. The modeling ap- proach presented in [9] is in this work exploited to design a simple but effective injection pressure control strategy to regulate the mean rail pressure. The control architecture is mainly composed by a model-based feed-forward controller coupled with an integral closed loop action. Parameters of both controllers are scheduled as function of engine speed and battery voltage to compensate model mismatch. The resulting control strategy is then experimentally tested to show its effectiveness. The paper is outlined as follows: in Sec. II the common rail system and its control oriented model are described for the sake of completeness, then an analysis on the disturbances affecting the injection pressure is drawn in Sec. III, finally, before conclusions, the design of the injection pressure controller and its experimental validation are reported in Secs. IV and VI, respectively. 2011 American Control Conference on O'Farrell Street, San Francisco, CA, USA June 29 - July 01, 2011 978-1-4577-0079-8/11/$26.00 ©2011 AACC 5273