The investigation reported in this thesis is concerned with the topic of soot formation and soot particle motion in the cylinder of a light duty automotive diesel engine. CFD has been employed to simulate in-cylinder conditions and to investigate the source of particles which are transferred to the oil. The accumulation of soot in the lubricating oil of diesel engines is one of the factors limiting the interval between oil changes and hence service interval. Soot particles can be transferred to oil film on the cylinder wall layers through the complex motion of the fluid flow in the cylinder. The paths of soot particles from specific in-cylinder locations and crank angle instants have been explored using the results for cylinder charge motion predicted by the Kiva-3v CFD code. Using the velocity fields from the simulation data, massless tracking of the in-cylinder soot particles in space and time is carried out employing a particle tracking with trilinear interpolation technique. From this investigation, new computational codes for the prediction of soot particle paths and soot particle size change along a specific path in a diesel engine have been developed. This investigation is the first numerical study into soot particle trajectories within an engine and thus opens up a novel branch of research of soot formation within internal combustion engines. Computed soot paths from the investigation show that soot particles formed just below the fuel spray axis inside the middle bowl area during early injection period are more likely sources of soot particles on the cylinder wall layers than those formed later. Soot particles that are formed above the fuel axis have less tendency to be transported to the cylinder wall layers thus are not likely to be the main source of soot at the cylinder walls. Soot particles that are from the bowl rim area are found to be another source of soot transfer to the boundary layer, as they are directly exposed to reverse squish motion during the expansion stroke. Soot particles that are formed near the cylinder jet axis during fuel injection tend to move into the bowl. These soot particles are found to be from the relatively less concentrated area. In contrast, particles from the most concentrated areas tend to be moving into the bowl and pose least risk of contaminating oil films on the liner. Sensitivity studies of soot particle paths to swirl show that engine operating with low swirl ratios are more vulnerable to soot in oil problem as low swirls cause the bulk fluid flow to be moving closer to the cylinder walls due to fuel jet velocity and reverse squish motions. Decreasing the spray angle lessens the possibilities of soot particles from being transported close the cylinder wall layers while increasing the spray angle increases the possibilities of soot from the bowl region to be transported close to the cylinder wall layers. The temporal and spatial evolution of soot particle size can be predicted by using the history of temperature, pressure and gas species along the paths. An explorative investigation has been carried out to determine the most suitable method to tackle this soot particle evolution. With proper multipliers, all approaches perform quite satisfactorily in terms of predicting the trend of size change. Soot particles that are likely to be transferred to the cylinder wall layers are predicted to change in size parallel to the average mass profile in the whole cylinder where they quickly peak to maximum at around 18° CA ATDC, and gradually decrease in size through EVO.

The crankshaft is a complex component, and as such, the influence of its geometric parameters on stresses seen under service loads is not well understood. The objectives of this work are to investigate the effects of a wide range of geometric parameters on stresses in overlapped crankshafts, to find correlation between results and to formulate simple methods of predicting peak stress levels: It is intended to achieve this by use of the Finite Element (FE) and Boundary Element (BE) methods. Individual crankthrows are loaded under the important load cases of bending and torsion. Stress concentration factors are determined by normalising peak stresses with respect to the nominal stress occurring in the most appropriate section in the neck between the fillets. Analyses are carried out in 2D and 3D, making use of symmetry as far as possible. Many of the governing dimensions of the crankthrow are included in the analyses; crankpin and journal diameters, crankpin and journal overlap, and web thickness. Variations in SCF are plotted over a wide range for each of these parameters. Additionally, features such as fillet size and shape, bore-holes, dimples, cut-back webs and oil holes are investigated. It is found that the effects on stress of individual parameter changes can be superimposed to accurately predict the effect of combining various parameter changes in one model. The crankpin and journal fillet radii and the length of the minimum section between the fillets are shown to be the critical parameters in determining the peak stress levels in the crankshaft. SCFs obtained from the range of analyses performed show good agreement with the classical theory of SCFs in notched bars. Bore-holes and dimples are found to offer significant benefits in terms of peak stress reduction, in addition to their common usage of reducing the out of balance crankpin mass. The FE and BE methods give accurate results for stress analysis of crankshafts and offer several advantages over traditional experimental techniques; they are ideally suited to parametric analyses, can be carried out relatively quickly, results are repeatable because boundary conditions can be exactly defined, and the cost of analysis is significantly reduced.

Cylinder pressure analysis is widely used in the experimental investigation of combustion processes within gasoline engines. A pressure record can be processed to reveal detail of charge burning, which is a good indicator of combustion quality. The thesis describes the evaluation of an approximate technique for calculating the mass fraction of the charge that has burnt; a novel approach for determining heat loss to the block; the development of a powerful system for combustion analysis; and the investigation of the correlation between the crank angle location of the 50% mass burnt and minimum timing advance necessary to obtain the maximum engine torque. A detailed examination has been carried out into the uncertainties in the determination of the mass fraction burnt as suggested by Rassweiler and Withrow. A revised procedure has been developed which does not require a priori identification of the combustion end point, and a new approach is suggested to calculate the polytropic indices necessary for the pressure processing. This particular implementation of the analysis is able to identify late burning and misfiring cycles, and then take appropriate steps to ensure their proper analysis. The problems associated with the assumption of uniform pressure; alignment of the pressure changes to the volume changes; pressure sampling rate; clearance volume estimation; and calibrating the acquired pressure to absolute are also evaluated. A novel method is developed to ascertain, directly from the pressure history, the heat loss to the cylinder block. Both experimental and simulated data are used to support the accuracy of the suggested heat loss evaluation, and the sensitivity of the method to its inputs is examined. The conversion of procedures for combustion analysis into a format suitable for undertaking high speed analysis is described. The analysis techniques were implemented so that the engine can be considered to be on-line to the analysis system. The system was entitled Quikburn. This system can process an unlimited number of cycles at a particular running condition, updating the screen every 1.5 seconds. The analysis system has been used to study the potentially beneficial correlation between the location of the 50% mass burnt and MBT. The correlation is examined in detail, and found to be valid except under lean fueling conditions, which is seen to be caused by slow flame initiation. It is suggested that the optimum location of the 50% mass burnt can be used as a reference setting for the ignition timing, and as an indicator of combustion chamber performance. An engine simulation was employed to verify that changes in bum shape account for the small variation seen in the optimum 50% bum locations at different operating conditions of the engine. The bum shape changes also account for the range of optimum locations of the 50% mass burnt encountered in different engines.

This thesis describes research undertaken to establish the advantages and disadvantages of using high pressure common rail fuel injection systems with multiple injection capabilities. The areas covered are detailed as follows. Oscillations in the rail pressure due to the opening of the injector can affect the quantity of fuel injected in subsequent injection events. The source of these oscillations has been investigated. A method of damping or reducing the oscillations has been defined and was applied. This successfully reduced the level of unpredictability of the quantity of injected fuel in subsequent injection events. A relationship between needle lift, injection pressure and the quantity of fuel injected was established. The effects of fuel injection parameters (main injection timing, split main separation and ratio) and engine operating parameters (boost pressure and EGR level) on emissions formations and fuel economy have been investigated at five operating points. Design of Experiments techniques were applied to investigate the effect of variables on pollutant emissions and fuel consumption. The sensitivity and linearity of responses to parameter changes have been analysed to assess the extent to which linear extrapolations will describe changes in smoke number (FSN) and oxides of nitrogen (NOx); and which parameters are the least constricting when it comes to adjustments of parameter settings on the FSN-NOx map. Comparing results for split main and single injection strategies at the five operating conditions shows that split main injection can be exploited to reduce NOx or FSN values at all conditions and both NOx and FSN simultaneously at high load conditions. The influence of changing engine speed and brake mean effective pressure (BMEP) on FSN and NOx emissions with given fixed values of parameter settings has been investigated. This established how much of the operating map could be covered by discrete calibration settings. Finally the variation in parameter settings required to maintain fixed FSN and NOx values across the operating map, near the optimum trade-off on the FSN-NOx map, was analysed. Combining the information gained from the individual investigations carried out highlighted some techniques that can be used to simplify the calibration task across the operating map, while also reducing the amount of experimental testing required.

The work described in this thesis concerns the continued development and application of a computational model to simulate the thermal behaviour of internal combustion engines. The model provides information on temperature and heat flow distributions within the engine structure, and on temperatures of oil, coolant and engine-out exhaust gas. Sub-models calculate friction levels, fuel flow rates and gas-side heat transfer, including the effects of exhaust gas recirculation (EGR), spark advance and turbocharging. The effects of auxiliary components such as a cabin heater, oil cooler, intercooler, supplementary heater and EGR cooler can also be simulated. Model developments are aligned towards improving the accessibility of the model and the scope of engine systems that can be simulated. Early versions of the model have been converted from 'C' into the current MATLAB/Simulink versions. The model structure and conversion process are described. New developments undertaken have focused on the external coolant circuit and include the modelling of the thermostat and radiator. A semi-empirical thermostat model is presented. A radiator model based on the effectiveness-NTU method is described. Simulations using the developed model, including the thermostat and radiator sub-models, investigate the effect of thermostat position on engine thermal behaviour. Positioning the thermostat on the inlet to the engine reduces thermal shock. Applications of the model to investigations of sensitivity and performance illustrate the accuracy of and confidence in model predictions. Assessments demonstrate that the model is relatively insensitive to variations of 100/0 in user inputs and is very sensitive to model assumptions if simulation conditions, implied in the model assumptions, are not matched to test conditions. A process for evaluating model performance is described. Evaluation exercises applied to three different engines demonstrate that values predicted by the model are to within 5 to 10% of experimental values. Investigations using the model of methods to improve warm-up times and fuel consumption prior to fully warm conditions show the benefits or otherwise of reduced thermal capacity, an oil cooler, a sump oil heater and an oil-exhaust gas heat exchanger. Each method is assessed over the New European Drive Cycle (NEDC) from a -10°C start. Of these methods, a combined reduction in coolant volume and engine structural mass is most beneficial for reducing coolant warm-up times. An oil-exhaust gas heat exchanger produces the greatest reduction in fuel consumption.

A systematic investigation of the effects of coolant-to-mainstream density ratio and mainstream acceleration on the heat transfer following injection through a row of holes in a flat plate into a turbulent boundary layer is described. A mass transfer technique was employed which uses a swollen polymer surface and laser holographic interferometry. The constant concentration of the test surface simulated isothermal conditions. Density ratios in excess of unity, representative of gas turbine operating conditions, were obtained using foreign gas injection into mainstream air. The experimental technique was validated for such measurements. The cooling film heat transfer coefficient was measured for a range of blowing configurations and flow conditions; the holes were spaced at three diameter intervals and inclined at 35° or 90° to the mainstream, and the ranges of the other pertinent test parameters covered were, 0.5 5 blowing rate 5 2.0, 1.0 5 density ratio S 1.52, and 0.0 S acceleration parameter S 5x 10'. However, the tests with mainstream acceleration were performed with 35° injection only. The heat transfer coefficient was found to be increased by injection, and with the blowing rate for both 35° and 90° injection. Close to the injection site, normal blowing produced higher heat transfer coefficients than angled blowing, but gave lower coefficients far downstream. There were large differences in behaviour between the two injection angles with varying density ratio. For normal injection, the heat transfer coefficient at a fixed blowing rate was insensitive to the variation of density ratio, whereas for 35° injection strong dependence was observed, an increase in the density ratio leading to a decrease in the coefficient. Similar behaviour for the inclined injection case was also found in the presence of strong favourable pressure gradient. As mainstream acceleration acts to suppress injection induced turbulence, the heat transfer coefficient under the film with and without density ratio was found to decrease in the presence of mainstream acceleration relative to that in absence of acceleration. The heat transfer coefficient was observed to relate to the acceleration parameter in an approximately linear manner, an increase in the acceleration resulting in a decrease in the coefficient. For normal injection, good scaling of the heat transfer coefficient including density ratios was achieved with the blowing parameter. For 35° injection, the coolant to mainstream velocity ratio was seen to scale the data best. Correlations for the heat transfer data using these scaling parameters. With these correlations data obtained at density ratios not representative of gas turbine practice can be adapted for design calculations. The predictions of a computational fluid dynamics general purpose program called PHOENICS were tested against the present measurements and those of others. In general, the computed results of film cooling effectiveness agreed reasonably well with available experimental data. The ability to predict the heat transfer coefficient associated with film cooling was satisfactory for normal injection, but not as satisfactory for injection through 35° holes.

The cold start performance of automotive diesel engines is currently poor when compared to similar gasoline units. This thesis describes an experimental and theoretical investigation into the factors limiting diesel cold start behaviour. Studies have been carried out on IDI and DI designs of engine. Start behaviour has been characterised in terms of times taken to complete various stages of startup, engine speed variations and processes which affect these. Combustion and friction behaviour have been investigated in detail. Engine friction losses are dependent on temperature. During start-up these losses are relatively high compared to those when the engine is fully-warm. The work output from combustion is low at low speeds, and prone to a further deterioration at lower temperatures. Consequently, combustion output during cold cranking is initially insufficient to overcome frictional losses. The start times are extended by the need to keep the starter motor engaged until heat generated in the engine causes frictional losses to fall. Eventually, when combustion output is able to overcome friction without the aid of starter motor work, idle speed is reached. Changes to fuel injection and glowplug parameters have been used to achieve a limited improvement in low-temperature starting. Measurements of engine friction have been carried out to determine the influence of temperature and speed, and the relative contributions from each of the main component assemblies. At low temperatures, much of the friction originates in hydrodynamically lubricated components such as journal bearings, due to high oil viscosity at low temperature. Additionally, engine friction as rotation begins has been shown to be far higher than measured by conventional "steady-state" motoring tests (over twice the quasi-steady state friction at -200 e). This initially high friction transient decays towards the quasi-steady values throughout the start. For crankshaft bearings, a friction model has been developed for cold start-up through to fully warm engine conditions. The friction behaviour in the bearings is dependent on thermal conditions around the friction surfaces. Models for the starter system and blowby processes are also presented as part of a broader theoretical investigation to assess the impact of design changes on start quality.

The overall objectives of this thesis were to investigate the potential of the peroxyoxalate chemiluminescence (POOL) reaction for the quantitative detection of target analytes in non-aqueous matrices and to compare quantitative performance with fluorescence detection. The target analytes investigated were polycyclic aromatic hydrocarbons (PAHs) and aliphatic amines. These were selected as an important class of compounds in engine exhaust emissions and a detergent additive in diesel fuel respectively. Chapter one outlines the challenges of analysing petroleum products and engine exhaust emissions and discusses the potential of luminescence techniques, particularly chemiluminescence (CL), for the quantification of trace components. The chapter also reviews the technique of flow injection (FT) as a means of sample delivery for CL detection and as a potential technique for field deployment. Liquid chromatography techniques are described as a means of separation of complex matrices, e.g. fuels and engine exhaust particulates, in the laboratory prior to CL detection. The luminescence properties of several PAHs were investigated in Chapter Two. Optimum excitation and emission wavelengths for eleven PAHs in four different solvents were determined using a batch fluorescence technique. A FI approach was used to determine PAH concentrations using fluorescence and POCL detection. Two aryl oxalates; bis(2,4-dinitophenyl)oxalate and bis(2,4,6-trichlorophenyl)oxalate were compared for their suitability for PAH determinations and an investigation of the key variables (e.g. concentration of aryl oxalate and hydrogen peroxide, mobile phase composition and pH) affecting POCL was performed. Recommendations for the optimum conditions for the determination of PAHs by POCL detection were determined, A comparison between a photodiode based detection device and a low power (12V) photomultiplier tube was also described. In Chapter Three the procedure of using POCL detection as a post column liquid chromatography (LC) detector for PAHs has been considered. The performance of the POCL detection system was compared with wavelength programmed fluorescence. Both reversed and normal phase LC was investigated and the suitability of POCL detection with each approach was discussed. Additionally the procedure for the LC separation and analysis of SRM 1649 (Urban Dust/Organics) and SRM 1650 (Diesel Particulate Matter) was described. The relative performance of fluorescence and CL detection are discussed. Chapter four describes the principles of multivariate calibration of spectrophotometric data, and three commonly applied techniques (PCR, PLSI and PLS2). Fluorescence data was obtained for synthetic mixtures of PAHs containing two, three, four and five components. A procedure whereby individual spectra were 'glued' together before undergoing data analysis has been developed and the results obtained discussed. POCL emission spectra for five PAHs were acquired using a two-dimensional charge coupled device (CCD). The sensitivity of the CCD system toward POCL detection of PAHs and a multivariate investigation using benzo[a]pyrene and benzo[k]fluoranthene has been described. The potential of the fluorescence and CL approaches used has been discussed. Chapter five describes the aryl oxalate sulphorhodamine-101 CL reaction and its application to the determination of amines. A FI optimisation of the reaction parameters is presented together with some quantitative data for the detection of a homologous series of amines and dodecylamine (a commonly added detergent compound in diesel fuels). The application of the technique toward the detection of dodecylamine in a diesel fuel matrix and the potential as a field deployable technique was also considered.

Compatibility of fuel delivery materials (FDM) with biodiesel fuel in the fuel delivery system (FDS) under real-life common rail diesel engine (CRDE) operation poses a challenge to researchers and engine manufacturers alike. Although standard methods such as ASTM G31 and ASTM D471 for metals and elastomers, respectively, are deemed suitable for evaluating the effects of water content, total acid number (TAN) and oxidation products in biodiesel on FDM degradation, they do not resemble the actual engine operation conditions such as varying fuel pressure/temperature as well as the presence of a wide range of materials in the FDS of a diesel engine. Hence, the current allowable maximum 20 vol% of biodiesel with 80 vol% of diesel (B20) for use in diesel engines to date is debatable. Additionally, biodiesel utilization beyond B20 is essential to combat declining air quality and to reduce the dependence on fuel imports. This thesis aims to elucidate the actual compatibility present between FDM and biodiesel in the FDS under real-life CRDE operation. This was achieved through multi-faceted experimentations which commenced with analyses on the deteriorated palm biodiesel samples collected during and after CRDE operation. Next, the fuel properties which should be emphasized based on the deteriorated fuel were determined. This was then followed by ascertaining the effects of the emphasized fuel properties towards FDM degradation. Ultimately, the actual compatibility of FDM with biodiesel under engine operation through modified immersion investigations was determined. FDM degradation acceleration factors such as oxidized biodiesel, TAN and water content were eliminated since these factors were not affected based on the analysed fuel samples collected after engine operation. No oxidation products such as aldehydes, ketones and carboxylic acids were detected while the TAN and water content were within 0.446% and 0.625% of their initial values, respectively. Instead, the biodiesel’s dissolved oxygen (DO) concentration and conductivity value were not only found to have changed during and after engine operation by -93% and 293%, respectively, but were also found to have influenced biodiesel deterioration under engine operation. These two properties were subsequently discovered to have adversely affected FDM degradation independently. The copper corrosion rate and nitrile rubber (NBR) volume change increased by 9% and 13%, respectively, due to 22% increase in the conductivity value. In contrast, the copper corrosion rate and NBR volume swelling reduced by 91% and 27%, respectively, due to 96% reduction in the DO concentration. Ultimately, copper corrosion and NBR degradation were determined to be lowered by up to 92% and 73%, respectively, under modified immersion as compared to typical immersion condition. These outcomes distinctly show that acceptable to good compatibility is present between FDM and biodiesel under CRDE operation. The good compatibility is strongly supported since only a maximum lifespan reduction of 1.5 years is predicted for metal exposed to biodiesel as compared to diesel for a typical component lifespan of 15 years. For the elastomers, acceptable compatibility is found present between elastomer and biodiesel based on the determined 11% volume change which conforms to the tolerance level of elastomer degradation as stated by the elastomer manufacturers. These are especially true for the evaluated metals and elastomers investigated under the modified laboratory immersion which replicates similar conditions to a real-life CRDE. Overall, this work has contributed to the advancement of knowledge and application of biodiesel use in diesel engines.

The deactivation of a cylinder on a 1.0litre three cylinder turbocharged gasoline engine has been investigated providing novel information on thermal and fuel consumption effects associated with the technology. This comes in light of providing solutions to reduce fuel consumption and CO2 emissions resulting from internal combustion engines. The investigation has been carried out through the PROgram for Modelling of Engine Thermal Systems (PROMETS). A version of PROMETS was extensively developed to characterise a commercially produced TCE not fitted with cylinder deactivation technology. Developments include an improved gas-side heat transfer expression to account for increased heat transfer to coolant due to the addition of an integrated exhaust manifold; addition of an expression to represent natural convection to model heating of quiescent coolant in the block; and a method to estimate the boosted intake manifold pressure past the throttle due to turbocharging on a gasoline engine. The 0-D approach used in this thesis compared to higher resolution computational tools has allowed for thermal and performance predictions to be made within a couple of minutes compared to several hours or days. In effect, PROMETS has been a time and cost effective tool during the development stages of a prototype engine. The PROMETS model indicated that no adverse changes in engine thermal behaviour arose with cylinder deactivation. The largest temperature change of < 400 occurs in the exhaust valve lower stem for the deactivated cylinder. Temperature changes in other components throughout the engine are an order of magnitude smaller. Although the largest temperature differences between the deactivated and firing cylinders were found to be in the range of < 70 , these remain within normal engine operating temperatures of < 100 . Also, by on-setting deactivation past an oil temperature of 40 , warm-up times were marginally extended compared to operation on all cylinders from key-on. Experimental inputs representing changes in engine gross indicated thermal efficiency and the work loss associated with the motoring of a piston complemented modelling work in predicting fuel consumption changes due to deactivation. Reductions in pumping losses account for the majority of the fuel consumption benefit associated with deactivating a cylinder. The main limitation in the employment of cylinder deactivation stems from the deterioration in the gross indicated thermal efficiency. Modelled results show that fuel consumption improvements are highest on low and part load operation envelopes. As such over the NEDC and FTP-75 benefits are in the range of 3.5%. Applying the technology over dynamically loaded cycles such as the WLTC and ARTEMIS, results in benefits of less than 1.6%. Further to modelling work on cylinder deactivation, experimental work has been carried out with the aim of allowing any engine size to be tested to cover transient drive cycles for future research. Future research could be in the aim of investigating technologies to reduce CO2 and emissions resulting from ICEs. Results show that the control solution implemented has allowed eddy-current dynamometers normally used for constant speed and brake load conditions to operate cycles such as the WLTC or any transient brake torque and engine speed pattern. Benchmark fuel consumption values for two engines of differing swept volume are within a 4g error band equivalent to a 0.36% and 0.67% percentage error band demonstrating the excellence of the control system.