Calibration of an Automotive Three-Way Catalyst Model Based on Synthetic Gas Bench Experiments

Nowadays, the Three-Way Catalyst (TWC) is the standard solution employed in the automotive industry to cut down CO, HC, and NOx tailpipe emissions of vehicles equipped with gasoline engines. The current trend towards more stringent regulations and severe testing procedures is pushing the design of modern propulsion and after-treatment systems to higher levels of complexity. This is especially true in the case of high-performance applications, for which compliance with the requirements of type approval is a serious task to accomplish. Therefore, it is necessary to develop numerical models embedding properly defined kinetic schemes, that can predict, robustly and accurately, the catalyst performances under various operating conditions. In fact, this is of paramount importance for effective optimization of the entire powertrain, with relevant savings in terms of both time and costs compared to experimental campaigns. Within this framework, this Master Thesis aims to build and calibrate a model of a TWC currently adopted by Ferrari, utilizing a one-dimensional (1D) multi-physics fluid-dynamic code, GT-SUITE. The required data set for the calibration of the global kinetic scheme, already defined by Ferrari according to the catalyst formulation, comes from a dedicated extensive experimental characterization performed at the Synthetic Gas Bench (SGB) by the REACT laboratory of the Queen's University of Belfast (QUB) on a catalyst sample extracted from the full-size monolith. The experiments includes isothermal OSC tests as well as light-off tests conducted under a temperature ramp from 100°C to 500°C and involving representative simple mixtures made of few selected species. This characterization methodology is mandatory to be able to separately analyze and calibrate the fundamental reactions occurring inside the catalyst, such as CO and HC oxidation and NO reduction, as well as oxygen storage on Cerium sites. In addition, also complex mixtures of synthetic gases are fed to the reactor to further investigate peculiar interactions, such as the effect of H2O on CO oxidation and the NO-C3H8 reaction. The tuning of the several kinetic parameters that define the chemical mechanisms, as well as the active site densities and the inhibition functions, is performed exploiting the Genetic Algorithm optimization tool embedded in the software. This tool turns out to be very effective in minimizing an objective function generally defined as the cumulative absolute error between the simulated and measured outlet concentration of the considered species. The calibrated model is finally validated over a light-off test involving a representative full mixture of the real exhaust gas at the inlet. The results show a substantial improvement in the catalyst performance predictivity with respect to the baseline kinetic scheme. The T50s, i.e. the temperatures at which each pollutant species reaches 50% of conversion efficiency, are much well correlated with the experimental evidence. However, limitations directly related to the lack of secondary oxidation and reduction pathways in the definition of the kinetic scheme emerge from the calibration activity. Future steps will then include the implementation of the reaction kinetics necessary to include also the N2O and NH3 formation, as well as the necessary modifications required to improve the OSC behavior.