Diagnostics of fuel injection systems in a CI engine fuelled with biodiesel based on vibration responses

In recent years, serious restrictions on diesel emission levels, combined with price instability and a significant increase in imports, has forced researchers to look for alternatives to this fossil fuel. Biodiesel is widely accepted as an alternative because it can be used in diesel engines without any substantial modifications and produced by sustainable resources. However, there are serious gaps in available knowledge regarding the effects of biodiesel blends on engine fuel injection systems and the engine combustion process. Therefore, this research focuses on the investigation into such effects through a vibration analysis of fuel injection systems in order to achieve nonintrusive quantitative diagnosis and hence condition monitoring of CI engines.

Having identified the specifics of technique gaps by a comprehensive literature study, this research firstly, investigates the dynamics of the fuel injection system with a CI engine running on biodiesel blends as fuels. This is achieved by numerical modelling analysis and experimental studies, which paves ways for using vibration response of fuel injection to diagnose the dynamic behaviour of different fuel properties. Then it investigates the of the change dynamic behaviour of fuel injection on engine combustion process. Finally, it explores the diagnostics of engine valve train clearance faults with an engine running with biodiesel and biodiesel blends based on engine fuel injection vibration responses.

A mathematical model has been developed and used to simulate the behaviour of the fuel injection system, including the fuel delivery and injector needle valve motions. It has concluded that the high pressure dynamic forces within the injection system will be affected by fuel properties such as fuel density, viscosity and bulk modules. The simulation results demonstrated; (i) that, the injector pressure is higher than that of the fuel injection pump, whose amplitudes are about 10% higher for biodiesels compared with petro-diesel; (ii) the levels of the pressure forces applied to the delivery valve and injector needle valve are also higher for biodiesel blends and (iii) nearly 1° (cam shaft) advance in the times of fuel injection rates and valve impacts with biodiesel and biodiesel blends. These predictions are confirmed by experimental results obtained by injection line pressures and pump vibrations and in-cylinder pressures.

Diesel engines are particularly prone to the engine combustion process primarily due to a fault in the fuel injection system and an abnormal clearance valve train conditions. The high-signal to noise ratio pump vibrations obtained from the pump body can be easily used for detecting and diagnosing faults from fuel injections. In the meantime, the research has also established that the pump vibration signals can be also used to recognise valve train diagnostics with medium effort of signal processing. It has found that the vibration levels become higher, due to the faults as a consequence of additional fuel supply to compromise the loss of overall power caused by poor combustion performance on the cylinder with an increased valve clearance. Moreover, B20 and B40 exhibit the similar changes with that of petro-diesel in the proposed high frequency envelop amplitudes (HFEA) whereas B100 shows less increased values. However, the pressure measurements are not very clear in representing these small changes in valve clearances for both the exhaust and inlet valves.

Compared with head vibration signals, which also can indicate the faults by a reduced level of vibration due to an effect combined reduced valve movement stroke with gas flow dampening, the pump vibration signals uniformly show the injection events and allow combustion uniformity between different cylinders to be diagnosed using a single transducer, whereas it may produce less accurate diagnosis by the head vibrations because of the close overlap of combustion and valve impact responses which needs complicated methods to be separated.