Emissions from gasoline engines remain an important issue worldwide as they are both harmful to health and contribute to green house effects especially under cold start conditions. A major challenge of the automotive industry is to reduce harmful emissions as much as possible whilst continuing to reduce CO2 emissions. Three-way-catalytic converters have been used very successfully to convert the harmful gases before release to the environment but these devices have to reach their light-off temperature in order to activate the chemical reactions. Therefore, the conversion time is delayed and during the pre light-off period, high levels of emissions are released. An investigation into methods capable of increasing catalyst temperature under cold start conditions has been carried out. The most beneficial technique used in this research was the secondary air method. The method introduced extra air into the exhaust manifold which allowed the engine to run rich and then the residual unburned fuel to be oxidised in the exhaust before approaching the converter. An experiment following a Box- Behnken design was used to study the effect of engine speed, spark angle, load, relative air/fuel ratio (lambda) and secondary air flow on pre-catalyst temperature. The study suggested the best result for the engine studied was to achieve fast catalytic light-off time was to run engine at 1225 rpm, spark angle of 0 degree BTDC, lambda of 0.82 and load of 0.5 bar BMEP. These settings allowed the remaining fuel to be burned with 5.87 kg/hr of secondary air in the exhaust manifold to achieve a pre-catalytic temperature of 631.1 QC and achieve light-off for all emissions within 17.2 seconds. The results were also used to build a temperature prediction model using the Matlab MBC toolbox and the best available model gave an R2 of 0.9997 by using radial base functions (RBF). However, the optimum conditions still produced cyclic variation in the combustion, giving an average COVimep of 14.8% during the pre-catalytic heating period which caused problems concerning engine smoothness. To derive a greater insight into the mechanisms governing the cyclic variability observed a simulation study was undertaken. The study used a simulation using Ricardo WAVE and Matlab Simulink to allow a detailed representation of some of the principle mechanisms giving rise to cyclic variability under cold start conditions. The study included combustion under rich and lean mixtures and considered the effect of variations of air/fuel ratios and residual gas fraction. As a result, the simulation showed a similar characteristic variability of heat release to that observed experimentally. The validation of the model for heat release showed that the predictions were under estimated by 0.49 % while under lean combustion, there was an under estimation of 2.07%. Both predictions had normally distributed residuals. The model suggested that the residual gas fractions were higher than the limit of 8.8% (under rich fuelling) or 8.0% (under lean fuelling) that was predicted to cause ignition delay to increase significantly and therefore contribute to high cyclic variability. ' An optimisation was carried out by varying camshaft angle in the simulation. The results suggest that retarding the exhaust camshaft position by 4 degrees (EVC 12 degrees BTDC) could reduce COVimep by 63.2% under rich combustion. In contrast, advancing the intake camshaft position suggested that the COVmep can be reduced but more experimental data is required to validate the results because variation of intake camshaft positions had a larger impact on pumping work than varying exhaust camshaft positions. These additional pumping losses result in higher air and fuel flow requirements. In summary, this thesis describes a detailed investigation into the effects of engine calibration on catalyst heating performance. One of the limiting factors in achieving rapid light-off is combustion variability. Extensions have been introduced to an industry standard ID engine simulation to allow realistic cyclic variability to represented and developed. These tools could allow cyclic variability to be considered more rigorously during a calibration exercise.
展开▼