High-speed flame imaging has been widely used to investigate flame propagation in optically accessible direct-injection spark-ignition (DISI) engines. Previous studies utilised a high-speed movie to measure the overall growth rate of the flame and to analyse the flame shape and its correspondence with engine performance and efficiency. This study proposes the flame image velocity (FIV), a new diagnostic method enabling time-resolved, two-dimensional flame front vector extraction and turbulence intensity calculation. The high-speed camera is used to record the propagating petrol flame and contrast variations are tracked to derive flow vectors. The PIVlab, aMatlab-based open-source code, is used for this flame front FIV analysis and the systematic optimisation of processing parameters is performed. The raw flame images are pre-processed using the contrast-limited adaptive histogram equalization (CLAHE) filter before a four-step Fourier transform (FFT) is applied. The interrogation window size for each step is optimised to achieve the highest flow vectors, which, for the studied cases, returns 84-84-24-24 pixels with a half overlap. A total of 100 combustion cycles are FIV processed for each test condition to tackle the inherent cyclic variations. The Reynolds decomposition is applied to individual cycles to derive highfrequency component magnitude, which is interpreted as turbulence intensity. A spatial filtering method is used for the decomposition with optimised cut-off lengths for minimal cyclic variations of the measured turbulence intensity. The new FIV method is proven useful in a case study using two injectors with different nozzle structures.The results show that a smaller hole diameter and counterbore hole shape leads to higher flame front vector magnitude, overall higher turbulence intensity and more uniform distribution of turbulence than the injector with a larger hole diameter and cylindrical hole shape. This FIV result explains higher engine power output and lower cyclic variations measured for the smaller hole injector.

The direct injection spark ignition (DISI) engines, while showing the extended controllability in operation conditions over port-injection engines, have a cyclic variation issue. For the past decade, many studies have been conducted to identify the causing factors of the cyclic variation and mitigate its negative effects. However, the fundamental knowledge of cyclic variation in DISI engines is still lacking. This experimental study measures cyclic variations in the in-cylinder pressure, early-stage burn duration and flame growth rate in a single-cylinder optical DISI engine and evaluates their correlations. While the in-cylinder pressure was recorded by the pressure sensor mounted in the spark plug, a high-speed camera was used to capture the flame propagation for over 100 individual cycles. The results show significant cyclic variations in both pressure derived data and flame image derived data; however, a clear positive correlation is observed between the burn duration and flame growth rate. Further analysis of the flames indicates higher level of flame front wrinkling measured in some individual cycles is a cause of the higher flame growth rate.