Preliminary analysis of a hydraulic variable valve actuation loss model for the control-oriented base engine calibration

Abstract

The achievement of the increasingly stringent emission limits for new passenger cars led to the development of complex engine architectures needed to perform advanced management strategies. Therefore, Variable Valve Actuation (VVA), Exhaust Gas Recirculation (EGR), Gasoline Direct Injection (GDI), turbocharging and powertrain hybridization have largely equipped modern internal combustion engines. Such complex systems provide a large set of degrees of freedom available for the engine regulation, introducing many new control variables within the control systems. The engine management software may contain hundreds of parameters that must be calibrated. Aim of the engine base calibration process is to identify proper values for the map, scalar and vector calibration parameters so that the values estimated by the functions are as close as possible to those that can be measured at the test bench in the same operating condition for the same quantity. Thousands of operating conditions can be analyzed and the physical quantities needed for the engine base calibration are acquired and recorded in a datasheet. This dataset is then used to calibrate the Engine Electronic Control Unit functions. The experimental activity, which can last several weeks, is the most critical stage of the process in terms of time and cost of implementation. Moreover, as the powertrain complexity increases, the calibration effort grows exponentially. In addition, the system complexity featured by current engines requires a further increase in experimental tests needed to achieve a reliable calibration. To overcome this criticality, authors, in previous papers, have proposed the adoption of a 0D-1D thermo-fluid dynamic simulation method used to generate mathematical and physical models of the engine behavior starting from a small subset of the engine operating conditions usually tested for calibration purposes. In particular, 0D-1D CFD simulation codes, after being reliably calibrated, could be useful to greatly reduce dynamometer tests of advanced powertrains from thousands of operating conditions to a few tens. Nowadays, in these models the losses are predicted by experimental correlation that need a calibration using experimental data. These are empirical non-predictive relationships, which are low sensitive to physical quantities’ change for the evaluation of Friction Mean Effective Pressure. To overcome this issue, the authors evaluate the possibility of using a specific numerical Physics-Based models be able to predict the engine frictions throughout its operating range. The authors propose to use these numerical Physics-Based models for replacing experimental correlation, that be coupled with 0D-1D CFD simulations, for generating complete engine Dataset starting from few experimental tests, in order to reduce experimental effort in Base calibration process. Specifically, Multibody 2D and 0D-1D CFD approaches are used to generate physical models of the engine hydraulic variable valve actuation system. The proposal core is based on a parametrical approach in order to analyze the model sensitivity related to specifics parameters that are more uncertain. The authors propose a specific automatic strategy for calibrating valvetrain numerical model, starting from a series of parameters highlighted by parametric analyses, for reducing uncertainty of the physical and geometric characteristics. In order to evaluate the quality of the model, its results are compared to experimental data of fourteen operating point: from first results, it is possible to ascertain that the model reproduce the analyzed phenomena at least for two order of magnitude. The physical model developed presents several circuits, including the electronic and hydraulic one. Moreover, the parameters identified were used for a base calibration, carried out by performing a vector optimization based on a Multi Objective Genetic Algorithm to calibrate the model. In order to understand the potential of this approach in the control applications, the optimization is applied only to 3 out of the 14 operating point. The calibrated parameters found were than applied to all 14 operating point, and the final average error related to the experimental data, in Friction Mean Effective Pressure (FMEP) terms, dropped to about 12%.The achievement of the increasingly stringent emission limits for new passenger cars led to the development of complex engine architectures needed to perform advanced management strategies. Therefore, Variable Valve Actuation (VVA), Exhaust Gas Recirculation (EGR), Gasoline Direct Injection (GDI), turbocharging and powertrain hybridization have largely equipped modern internal combustion engines. Such complex systems provide a large set of degrees of freedom available for the engine regulation, introducing many new control variables within the control systems. The engine management software may contain hundreds of parameters that must be calibrated. Aim of the engine base calibration process is to identify proper values for the map, scalar and vector calibration parameters so that the values estimated by the functions are as close as possible to those that can be measured at the test bench in the same operating condition for the same quantity. Thousands of operating conditions can be analyzed and the phys...