This thesis aims to contribute to the research and development of a new highly efficient split-cycle engine concept – the double compression-expansion engine (DCEE) – by expanding the knowledge of combustion processes suitable for this and, potentially, other modern engines, via experimental and computational studies. In this work, first, the importance of continued improvement of internal combustion engines is demonstrated by comparing the life-cycle CO2 emissions of different modes of transport, including walking and bicycling. Then, an isobaric combustion concept is proposed for use in modern high-pressure combustion engines, such as the DCEE. Isobaric combustion is compared to conventional diesel combustion at different pressure levels, fueling, and EGR rates, and shown to reduce cylinder wall heat transfer losses by 20 %, simultaneously improving the NOx emissions by a factor of two. An in-situ injection rate measurement technique is developed and applied to improve the understanding of the complex injection strategies required for isobaric combustion. It is also shown that isobaric combustion is possible to achieve with a single fuel injector, but using multiple injectors may offer additional benefits of even lower heat losses, better heat release control, and improved soot and NOx trade-off. Then, an alternative combustion system to the diesel isobaric is proposed – a hydrogen direct-injection (DI) compression-ignition (CI) combustion concept, which has the advantage of ideally eliminated CO2 and soot emissions. DICI H2 combustion is found to differ significantly from conventional diesel, most importantly, in terms of the injected and retained momentum, and in-cylinder flow patterns and fuel-air mixing. Thus, a completely different optimization path must be taken for H2 engines, which involves maximizing the free-jet mixing phase of combustion while minimizing the momentum-dominated global mixing phase. This is achieved computationally in this work by adapting the combustion chamber shape to the H2 jets and modifying the injector nozzle, which proved effective. Finally, hydrogen combustion is computationally compared to diesel in the context of the DCEE on the basis of thermodynamic system parameters and detailed energy breakdown, and proved superior. Brake thermal efficiencies in the range of 56 % are demonstrated for the entire DCEE powertrain fueled with hydrogen.

A swirl injector for a rocket engine being developed by \emph{AESIR} (Association of EngineeringStudents in Rocketry) was simulated with different geometric parameters. The swirl injector is usedto atomize the ethanol used as fuel and to create a spray that mixes well with the oxidizer withinthe combustion chamber. Inlet slot angle (90, 75, 60 and 45 degrees), swirl chamber length (15, 20and 25 mm) and outlet orifice diameter (3, 6 and 9 mm) were examined.Previous studies in swirl injectors show that CFD can be used to analyze the flow in such aninjector, furthermore theoretical models exist that can predict some of the general characteristicsof the flow. Previous studies have also simulated transient behavior and flow features effectingbreakup of fuel flowing through a swirl injector.A steady state simulation using Volume of Fluid (VOF) multiphase modeling and $k$-$\omega$ \emph{SST}turbulence modeling was used to simulate the swirl injector intended for the rocket engine. It wasfound that a wider outlet orifice would give a wider cone angle of spray. This is desirable in thecurrent rocket engine design as it will promote greater mixing of fuel and oxidizer higher up in thecombustion chamber. No large variances was observed when different inlet slot angles was simulated. Ashorter swirl chamber length reduced the amount of losses in energy due to viscous forces. The flowafter the outlet orifice was not simulated so the effect of turbulence kinetic energy and energylosses outside of the swirl injector have not been analyzed, previous studies have indicated thatturbulent kinetic energy does have an effect on the breakup and atomization of the fuel.It was concluded that using a wider outlet orifice of 9 mm gave the best results out of the differentgeometric parameters analyzed and the swirl chamber length should be a short as possible.

Low-emissions combustor design is crucially important to gas turbine engine manufacturers. Unfortunately, many designs are susceptible to unsteady oscillations that can result in structural fatigue and increased noise. Computational approaches that resolve flow unsteadiness, for example Large Eddy Simulation (LES), are being explored as one avenue to help understand such phenomena. However, in order to quantifY the accuracy of LES predictions, benchmark validation data in suitably chosen test cases are required. Comprehensive experimental data covering both time-averaged and timeresolved features are currently scarce. It was the aim of this thesis, therefore, to provide such data .in a configuration representing the near-field of a typical gas turbine fuel injector. It was decided to focus on the fuel injector since many unsteady events are believed to originate because of the transient interactions between the fuel injector flow and the main combustor flow. A radial fed two-stream fuel injector, based on a preexisting industrial gas-turbine Turbomeca design was used, since this geometry was known to be susceptible to unsteadiness. The fuel injector was investigated under isothermal conditions to place emphasis on the fluid mechanical behaviour of the fuel injector, including detailed capture of any unsteady phenomena present. Light Sheet Imaging (LSI) systems were used as the primary experimental technique to provide high quality spatially and temporally resolved instantaneous velocity and scalar field information in 2D planes (using ParticieImage Velocimetry (PIV) and Planar LaserInduced Fluorescence (PUF) techniques). Several methods were employed to extract information quantifYing the flow unsteadiness and improve visualisation of timedependent large-scale turbulent structures. Proper Orthogonal Decomposition (POD) analysis enabled clear identification of the dominant modes of energy containing structures. The results indicated that periodic high-energy containing vortex structures occurred in the swirl stream shear layer, emerging from the fuel injector. These formed a two-strong two-weak rotating vortex pattern which propagated down the main duct flow path. The formation of these vortices was found to be a function of the swirl number and originated due to an interaction between the forward moving swirl flow and the furthest upstream penetration point ofthe recirculation zone present in the main duct flow. Dependent on the magnitude of the swirl number (influencing the swirl stream cone angle) and the geometry of the fuel injector, the vortex formation point was sometimes found inside the fuel injector itself. If the vortices originated inside the fuel injector they appeared much more coherent in space and time and of higher energy. A second unsteady high energy containing phenomenon was also identified, namely a Precessing Vortex Core (PVC), which was damped out if the fuel injector contained a central jet. The dynamics of the PVC interacted with the dynamics of the swirl stream shear layer vortices to reduce there strength. Transient scalar measurements indicated that there was a clear connection between the unsteady vortex pattern and the rate of mixing, resulting in bursts of high heat release and is therefore identified as one source of combustor oscillations. Future fuel injector designs need to pay close attention to these unsteady features in selecting swirl number and internal geometry parameters.

Heavy-duty trucks are important links in the logistic chains of transport. Critical components in trucks include fuel injectors in which inaccuracies can lead to severe financial damage and higher emissions. Intelligent and efficient ways to detect such scenarios are thus of high importance. This thesis applies machine learning algorithms to measured or estimated engine data, focused on gas exchange signals, to detect inaccuracies in fueling quantities. The fueling inaccuracies considered were of low deviations from the nominal curve, with magnitudes not covered by the currently used fueling diagnostics. The data used for the models was generated from Scania test cell engines where different setups of injectors were deliberately set to over- or underfuel.  Seven different machine learning models were used on the data and evaluated on how well they could detect deviations from nominal fueling. The tests were mainly done with a pure data-driven approach but also improved through different data selection techniques and using domain knowledge. An investigation to connect the findings within the thesis to real customer data was initiated in order to make the results useful for e.g. predictive maintenance. The complications connected to why this was not ultimately achieved were discussed.

The diesel engine in heavy-duty trucks is a complex system with many components working together, and a malfunction in any of these components can impact engine performance and result in increased emissions. Fault detection and diagnosis have therefore become essential in modern vehicles, ensuring optimal performance and compliance with progressively stricter legal requirements. One of the most common faults in a diesel engineis faulty injectors, which can lead to fluctuations in the amount of fuel injected. Detecting these issues is crucial, prompting a growing interest in exploring additional signals beyond the currently used signal to enhance the performance and robustness of diagnosing this fault. In this work, an investigation was conducted to identify signals that correlate with faulty injectors causing over- and underfueling. It was found that the NOx, O2, and exhaust pressure signals are sensitive to this fault and could potentially serve as additional diagnostic signals. With these signals, two different diagnostic methods were evaluated to assess their effectiveness in detecting injector faults. The methods evaluated were data-driven residuals and Random Forest classifier. The data-driven residuals, when combined with the CUSUM algorithm, demonstrated promising results in detecting faulty injectors. The O2 signal proved effective in identifying both fault instances, while NOx and exhaust pressure were more effective at detecting overfueling. The Random Forest classifier also showed good performance in detecting both over- and underfueling. However, it was observed that using a classifier requires more extensive data preprocessing. Two preprocessing methods were employed: integrating previous measurements and calculating statistical measures over a defined time span. Both methods showed promising results, with the latter proving to be the better choice. Additionally, the generalization capabilities of these methods across different operating conditions were evaluated. It was demonstrated thatthe data-driven residuals yielded better results compared to the classifier, which requiredtraining on new cases to perform effectively.

Current lean-premixed fuel injector designs have shown great potential in terms of reducing emissions of pollutants, but such designs are susceptible to combustion instabilities in which aerodynamic instability plays a major role and also has an effect on mixing of air and fuel. In comparison to prototype testing with combustors running in operating conditions, computational approaches such as Large Eddy Simulations (LES) offer a much more cost-effective alternative in the design stage. However, computational models employed by LES require validation by experimental data. This is one of the main motivations behind the present experimental study. Combined particle image velocimetry (PIV) and planar laser induced fluorescence (PLIF) instrumentation allowed simultaneous measurements of velocity vector and a conserved scalar introduced into the fuel stream. The results show that the inner swirl shear layer features two pairs of vortices, which draw high concentration fuel mixture from the central jet into the swirl stream and causes it to rotate in their wakes. Such periodic entrainment also occurs with the characteristic frequencies of the vortices. This has clear implications for temporal variations in fuel/air ratio in a combusting flow; these bursts of mixing, and hence heat release, could be a possible cause of mixing-induced pressure oscillation in combusting tests. For the first time in such a flow, all 3 components of the turbulent scalar flux were available for validation of LES-based predictions. A careful assessment of experimental errors, particularly the error associated with spatial filtering, was carried out. Comparison of LES predictions with experimental data showed very good agreement for both 1st and 2nd moment statistics, as well as spectra and scalar pdfs. It is particularly noteworthy that comparison between LES computed and measured scalar fluxes was very good; this represents successful validation of the simple (constant Schmidt number) SGS model used for this complex and practically important fuel injector flow. In addition to providing benchmark data for the validation of LES predictions, a new experimental technique has been developed that is capable of providing spatially resolved residence time data. Residence times of combustors have commonly been used to help understand NOx emissions and can also contribute to combustion instabilities. Both the time mean velocity and turbulence fields are important to the residence time, but determining the residence time via analysis of a measured velocity field is difficult due to the inherent unsteadiness and the three dimensional nature of a high-Re swirling flow. A more direct approach to measure residence time is reported here that examines the dynamic response of fuel concentration to a sudden cutoff in the fuel injection. Residence time measurement was mainly taken using a time-resolved PLIF technique, but a second camera for PIV was added to check that the step change does not alter the velocity field and the spectral content of the coherent structures. Characteristic timescales evaluated from the measurements are referred to as convection and half-life times: The former describes the time delay from a fuel injector exit reference point to a downstream point of interest, and the latter describes the rate of decay once the effect of the reduced scalar concentration at the injection source has been transported to the point of interest. Residence time is often defined as the time taken for a conserved scalar to reduce to half its initial value after injection is stopped: this is equivalent to the sum of the convection time and the half-life values. The technique was applied to a high-swirl fuel injector typical of that found in combustor applications. Two test cases have been studied: with central jet (with-jet) and without central jet (no-jet). It was found that the relatively unstable central recirculation zone of the no-jet case resulted in increased transport of fuel into the central region that is dominated by a precessing vortex core, where long half-life times are also found. Based on this, it was inferred that the no-jet case may be more prone to NOx production. The technique is described here for a single-phase isothermal flow field, but with consideration, it could be extended to studying reacting flows to provide more insight into important mixing phenomena and relevant timescales.