Ion currents in diesel engines

This thesis documents an experimental and modelling investigation into ion formation in diesel engines, its uses in the field of engine performance and emissions prediction and the mechanisms by which these uses are made possible. Ion sensors have been employed in engines for a variety of purposes, including estimation of air-fuel ratio, start of combustion and in-cylinder pressure, detection of knock, misfire and combustion resonance, prediction of soot formation, and control of spark advance timing and start of injection. This study presents an experimental setup to measure ion currents in a direct-injection diesel engine and shows further uses of the ion current in engines, including prediction of engine-out torque, peak heat release rate, cumulative heat release and NOx emissions via the use of statistical correlations. Whilst these correlations are simple to obtain, satisfactory explanations for their existence have not yet been put forward. In-depth analysis of the relationships between the ion current and several species of importance to heat release and emissions formation has been inadequate. In contrast, ions in flames have been extensively studied by chemical kineticists for decades. The identities of the most abundant ions in flames and the mechanisms for their formation and neutralisation are well known for laminar flames at atmospheric or lower pressure. The physics governing charge transport in gases which is responsible for the development of an ion current from a concentration of charged species is also well established. This study brings together the two fields of engine research and chemical kinetics in order to provide an understanding of the chemistry underlying the correlations seen previously. Combustion systems of gradually increasing complexity are studied in order to model ion current trends in engines and establish causal links between the ion current and various measures of engine performance or emissions. Initially, a simple and well understood system is considered; methane combustion with ion formation is studied in a zero-dimensional model using a perfectly stirred reactor burning a methane-air mixture via the use of a well-known chemical kinetic mechanism augmented with a simplified ionic sub-mechanism. This mechanism is reduced to produce a highly simplified mechanism capable of predicting the ion current for high-pressure methane-air flames at near-stoichiometric mixture fractions. Comparisons of the performance of this mechanism to those of two detailed well-established methane combustion mechanisms show very similar results. A 1-D laminar flame code is used to validate temperature distribution, major species distributions and the laminar flame speed against established experimental data. Validation of this mechanism is performed in a 2-D CFD simulation of the flame propagation, heat release and ion current developed during the constant-volume combustion of methane in air at different elevated values of initial chamber pressure and near-stoichiometric equivalence ratios. These results predict the same ion current trends as seen in the experiment and give some insight into the relationship between heat release and the ion current in a simple combustion event. The effects of load, speed or other engine operating parameters on the ion current are difficult to predict when using simple models such as zero-dimensional or constant-volume models. Ion formation in sprays of diesel fuel is therefore studied using a phenomenological diesel engine model with an n-heptane oxidation mechanism to model heat release and species formation. The n-heptane mechanism is augmented with an ionic sub-mechanism based on the ionic methane mechanism developed in this study, and with a reduced NOx mechanism. The modelled in-cylinder pressure is found to agree well with experimental measurements. The modelled ion current displays the same trends as the measured signal but soot formation is not modelled, leading to inaccuracy in the cycle-minimum value of the ion current. The reasons for the correlations presented here are investigated in zero-dimensional models and in the phenomenological engine model. Sensitivity analyses are used to identify species important to both heat release and electron production. Heat release is found to be correlated with electron formation due to the temperature dependence of the H-O system radicals, in particular that of OH. A similar study is conducted for NOx emissions. The NOx formation rate is also found to be correlated with electrons due to the H-O system radicals, via reactions involving the O radical. The formations of both species in a spray from a diesel injector in an engine are modelled using the same phenomenological model; predictions of their locations in the spray were found to match previous experimental results. Further uses of the ion sensor in engines are briefly discussed --- those of soot prediction and engine diagnostics for conditions leading to increased soot emissions. Soot is found to affect the sensor response primarily via the deposition of an electrically conductive layer on the sensor and the consequent alteration of the sensor resistance. The ion sensor is found to be able to detect two conditions leading to increased soot emissions, worn injectors and increased backpressure in the exhaust system. Detailed studies into sensor geometry and faulty engine operation are required for these applications to be robust, which are beyond the scope of this study but point out a possible route future ion current studies may take.