Search

Selective catalytic reduction (SCR), based on the injection of urea-water-solution (UWS), is one of the prevailing and more effective approaches to reduce NOx emissions in diesel engines. To improve the performance and durability of the system, it is crucial to develop reliable simulation tools able to correctly describe not only the urea conversion into ammonia and the mixing with exhaust gases but also the possible formation of solid deposits along with the engine’s exhaust line.

Virtualization of powertrain components allows the front-loading of conventional vehicle calibration and validation tasks to Model-in-the-Loop (MiL) and Hardware-in-the-Loop (HiL) simulations. This approach is based on the utilization of highly accurate physics-based powertrain models that enable a seamless system validation using virtual testing methods in order to ensure cost-effective powertrain development by reducing hardware tests. Proper modelling methods target the optimum between parametrization effort, model accuracy and required computing power to grant the real-time (RT) capability of the simulation, which is mandatory for HiL simulation. In this paper two validated modelling approaches and their implementation into a MiL environment are introduced and discussed. The approaches are the MATLAB/Simulink based Mean Value Engine Model (MVEM) and the Fast-Running Modelling (FRM) of GT-Power. After the models integration in a Simulink frame, the responses of a model-based control unit with the two simulation models were evaluated using real experimental data. In transient cycles, the controller showed a different reaction to the feedback signals of the two engine models. The purposes of the conducted investigation are mainly to evaluate strong and weak points of both approaches and to propose the best-practice modelling approaches for virtual calibration and validation. A comparative rating shows the main advantage of the MVEM in the flexibility for HiL-based systems and the model training effort for the FRM.

•Reinforcement Learning algorithm used to optimize the energy management of a PHEV•The strategy achieves performance close to the Dynamic Programming benchmark•Strategy robustness assessed on a wide range of real-world driving conditions•Impact of different reward functions on the performance of the algorithm

To comply with stringent NOxemission regulations, automotive diesel engines require advanced aftertreatment catalytic systems, such as lean NOxtraps (LNTs). Considering that test bench and chassis dyno experimental campaigns are costly and require a vast use of resources for the generation of data; therefore, reliable and computationally efficient simulation models are essential in order to identify the most promising technology mix to satisfy emission regulations. In the literature, a large number of simulation models for LNT kinetics can be found, realized for laboratory-scale samples and validated over synthetic gas bench (SGB) experimental tests, while full-size models validated over engine-dyno driving cycle data, crucial for industrial applications, are missing. In the current work, a simulation model of an LNT device is built to predict NOxstorage and reduction, starting from SGB laboratory tests and finally validated over driving cycle data. The experiments including light-off, NOxstorage and reduction (NSR), and oxygen storage capacity (OSC) characterization, were performed on a laboratory-scale sample extracted from a full-scale monolith. Light-off tests have been conducted under a temperature ramp cycle from 120 °C to 380 °C, while OSC and NSR tests were performed under isothermal conditions at five temperature levels, ranging from 150 °C to 400 °C. OSC tests were performed to characterize oxygen storage capacity of ceria sites and water gas shift (WGS) reaction over the precious metals by controlling inlet species concentrations with periodic lean/rich pulses. NSR experiments were then performed by alternating a lean inlet composition to reproduce adsorption/desorption of NOxwith a rich inlet composition feed with three reductants (H2, CO, and C3H6) to replicate NOxreduction reactions. A global kinetic scheme was defined by means of a one-dimensional (1D) engine simulation fluid-dynamic code, GT-SUITE, to model oxidation reactions (CO, HC, NO), NOxadsorption/desorption, oxygen storage and NOxreduction reactions. The kinetic parameters were obtained using Arrhenius plots with the aim to minimize the error between simulated and experimental NOx, reductants, N2O and NH3concentrations, reaching a satisfactory agreement with measurements.

•Deep neural networks optimize EMS of HEV through driving pattern recognition.•Real-time capable control strategy.•Strategy robustness assessed on real-world driving cycles.•Strategy achieves sub-optimal fuel consumption.

•Deep Recurrent Neural Networks used to optimize the energy management of a PHEV.•The strategy achieves performance close to the Dynamic Programming benchmark.•Strategy robustness assessed on a wide range of real-world driving conditions.

This paper presents an original model-based approach to the system level development and calibration of diesel powertrains for passenger cars, implemented in a comprehensive software tool. The models for the different subsystems are coupled to execute optimisation loops, involving also the system calibration, and are characterised by flexibility of usage, limited tuning effort and reduced computational time. As an example of the general approach, the engine model is described and some results from its execution are shown and compared with the measurements. Finally, a practical example is given of the usage of the tool for system level optimisation through a specifically developed methodology.

Nowadays, electric vehicles represent a promising solution for reducing the fuel consumption (and thus the carbon dioxide emission) and the pollutant emissions of road vehicles, especially in highly congested urban areas, although their driving range and usability still limit the customer acceptance even for a city car application. Extended-range electric vehicles may partly overcome these limitations, having an auxiliary power unit which can provide electrical energy to the powertrain once the battery has been depleted. On the other hand, the operations of such an auxiliary power unit should be almost completely unnoticeable in order to avoid any impacts on the electric driving experience, resulting in considerably reduced drive-related noise. Therefore the aim of this work is the design, through numerical simulation, of a powertrain controller capable of minimizing the carbon dioxide emission of a range-extended electric vehicle and, at the same time, avoiding any discomfort for the passenger related to the auxiliary power unit operations. Starting from the development of a powertrain controller focused on minimization of the carbon dioxide emission of the vehicle, the main noise targets will be defined and their effects on the energy management system will be analysed.

Image, graphical abstract