Virtual calibration of an ICE eco-powertrain and study of NVH implications on the combustion noise through sound quality analysis and fast auralization methods.

The objective of this thesis work is to develop a fast-running virtual engine combustion model, with the aim of estimating both performance and NVH indicators. A predictive model enables the forecast of the burn rate and, consequently, the exploration of the effects of calibration parameters on both performance and sound quality. This virtual engine model can generate pressure traces for all four cylinders, which can be easily processed for NVH implications. The model has been validated and calibrated using experimentally acquired measurements. A set of calibration settings have been varied during steady-state and run-up tests. During these tests, pressure traces and engine noise (captured through microphones) were also recorded. To isolate the combustion noise from the signals measured by the microphones, transmissibility function between the combustion chamber pressure and the acoustic pressure at the microphones have been calculated using the Wiener coherence method. These transmissibility functions were then used to create a modal model through Polymax algorithm and subsequently a state-space model. It was possible to replicate the noise and to conduct a sound quality analysis, comparing the results obtained from the virtual model with the experimental outcomes. Both the combustion model and the state-space model operate in the time domain, and they are also computationally efficient, allowing them to provide real-time results. This feature makes them well-suited for real-time applications.