The GDI engine has a number of practical advantages over the more traditional port-fuel injection strategy, however a number of challenges remain the subject of continued research in an attempt to fully exploit the advantages of the GDI engine. These include complex in-cylinder flow fields and fuel-air mixing strategies, and significant temporal variation, both through an engine cycle and on a cycle-by-cycle basis. Despite advances in experimental techniques, the relative difficulty and cost of taking detailed measurements remains high, thus computational techniques are an integral part of research activities. The research work presented in this thesis has focused on the use of detailed 3D-CFD techniques for investigating physical phenomena of the in-cylinder flow field and fuel injection process in a single cylinder GDI engine with early injection event. A detailed validation of the numerical predictions of the in-cylinder flow field using both the RANS RNG k-ε turbulence model and the Smagorinsky LES SGS turbulence model was completed with both models showing good agreement against available experimental results. A detailed validation of the numerical predictions of the fuel injection process using a Lagrangian DDM and both RANS RNG k-ε turbulence model and Smagorinsky LES SGS turbulence model was completed with both models showing excellent agreement against experimental data. The model was then used to investigate the in-cylinder flow field and fuel injection process including research into: the three dimensional nature of the flow field; intake valve jet flapping, characterisation, causality and CCV, and whether it could account for CCV of the mixture field at spark timing; the anisotropic characteristics of the flow field using both the fluctuating velocity and turbulence intensity, including the increase in anisotropy due to the fuel injection event; the use of POD for quantitatively analysing the in-cylinder flow field; investigations into the intake valve, cylinder liner and piston crown spray plume impingement processes, including the use of a multi-component fuel surrogate and CCV of the formed liquid film; characterisation and CCV of the mixture field though the intake and compression strokes up to spark timing. Finally, the predicted turbulence characteristics were used to evaluate the resultant premixed turbulent combustion event using combustion regime diagrams.

Direct fuel injection technology is increasingly being applied to the spark ignition internal combustion engine as one of the many actions required to reduce the CO2 emissions from road transport. Whilst the potential for CO2 reductions is compelling, the technology is not without disadvantages. Early examples typically emitted over an order of magnitude more Particulate Matter (PM) than vehicles with conventional spark ignition engines. Consequently, future revisions to European and North American exhaust emissions legislation are likely to regulate the particulate emissions from vehicles with direct injection gasoline engines. This thesis undertakes to investigate a) instrumentation capable of simultaneously resolving the number concentration and size distribution of particles in the 5-1000 nm size range and b) the factors affecting the PM emissions from spark ignition engines with direct fuel injection. The first objective is achieved by evaluation and comparison of a differential mobility spectrometer; photo-acoustic soot sensor; condensation particle counter and electrical low pressure impactor. To address the second question, a differential mobility spectrometer is applied to quantify the PM emissions from a number of direct injection gasoline engines, together with investigation of their dependence on various calibratable parameters, operating temperature and fuel composition. The differential mobility spectrometer showed good agreement with the other more established instruments tested. Moreover, it exhibited a faster time response and finer resolution in particle size. The number weighted size distribution of the PM emitted was typically lognormal with either one or two modes located between 20 and 100 nm. Chemical analysis of PM samples showed the presence of elemental carbon, volatile organic material and sulphates. Transient PM measurements enabled short time-scale events such as mode switching between homogeneous and stratified mixture preparation to be identified. PM number concentrations in stratified mode exceeded those in homogeneous mode by a factor of 10-100. Dynamometer based experiments showed that PM emissions increase for rich air fuel ratios, retarded fuel injection and advanced ignition events. They also demonstrated a strong dependence on fuel composition: the highest PM emissions were measured with an aromatic fuel, whereas blending alcohols such as methanol or ethanol tended to suppress PM emissions, particularly in the accumulation mode size range. These measurements are amongst the first of their kind and demonstrate the applicability of the differential mobility spectrometer to the measurement of ultra-fine particulate emissions from engines with direct fuel injection systems. Numerous explanations are put forward to describe the data obtained, together with suggestions for future work on PM control and abatement.

No description available.