| Nowadays, many techniques are developed in an attempt to achieve reliable,efficient and clean combustion for internal combustion engines. However, strongin-cylinder cycle-to-cycle variations (CCV) prohibited these technologies from reachingtheir full potential. In this study, to visually and systematically investigate the CCV ofin-cylinder processes, an optical single-cylinder spark-ignition direct-injection (SIDI)engine was designed and built. Several laser diagnostics techniques were speciallydesigned and optimized to systematically study the CCV of fuel spray, in-cylinder airmotion, flame kernel formation and the flame propagation. High-speed fuel sprayimaging was conducted in both the constant-volume chamber and the optical engine.Time-resolved particle image velocimetry (TRPIV) system are optimized to accuratelymeasure the in-cylinder air motion during the intake and compression strokes forhundreds of consecutive cycles. The flame kernel formation and flame propagation ofmultiple cycles are investigated by simultaneously recording in-cylinder flame imagesand in-cylinder pressure in kilohertz frequency for multiple cycles. The high recordingrate allows the computation of early flame propagating velocity fields by the noveltechnique based on the cross correlation of flame patterns. The uncertainty of theobtained vectors is evaluated to be less than5%.To extract useful information from the complex and large amount of in-cylinderexperimental data, the proper orthogonal decomposition (POD) technique was developedfor both the scale fields and flow fields. For the past decade, POD has gained increasingapplication for quantitatively comparison of simulation and experiment data, as well asthe identification of varying flow structures, especially the cyclic variation in the ICengines. However, the application can be very perplexing, and physical interpretation ofPOD is ambiguous. The flow patterns observed in the POD modes (or basis functions) are sometimes described at coherent structures, but the justification of this is not obviousfrom the mathematical procedures. In this study, the basic properties of POD areunambiguously identified for the first time by the following steps. Firstly, the principleand mathematical procedure of POD were detailedly described. Secondly, as applicationof POD can be daunting, synthetic velocity fields were used here to reveal the PODproperties rather than utilizing the real stochastic in-cylinder flow data. The syntheticflow fields were well designed to isolate the different types of CCV, and then conductingthe POD on these flow fields can intuitively and unambiguously reveal the properties ofPOD. Thirdly, two sets of extreme in-cylinder flow fields were employed to furtherillustrate how to use POD in engine research.POD can be performed on scalar fields as well. In this study, POD was applied onspray stability analysis for the first time. It was demonstrated that the POD can be apowerful and elegant tool for spray variation study. As the key component, fuel injectorplays a vital role in SIDI engines. Therefore, to examine the effect of fuel injectionsystem on the spray variations, and remove the influence from the in-cylinder air flow,the spray variations are investigated by performing the POD on the spray data sets fromthe constant-volume chamber. The results reveal that the fuel injection pressure, ambientpressure, fuel temperature and fuel properties have subtle effect on spray variation.Overall, the CCV of spray under quiescent ambient air condition is negligible.As the next step, the air motion is measured in the optical engine in motor modewithout fuel spray to remove the effect from the fuel spray, and performing POD onlyserves to study the in-cylinder air motion stability. A relatively novel technique, namelyphase-invariant POD, is used with the purpose to simultaneously explore the flowevolution and the cyclic variations. Instead of performing the POD on the velocity fieldsof different cycles at a fixed crank-angle degree (CAD), phase-invariant POD performsPOD on the velocity fields from different phases (i.e. CADs) of multiple cycles. In thismanner, a single set of POD modes was obtained, and can be used to represent any phaseof the flow. The changes of POD coefficients over phases illustrate the flow evolutionintra cycle. Moreover, the coefficients from different cycles for the same CAD quantifythe variations between cycles. Only the first three POD modes can capture more than80%of the total kinetic energy, and the flow evolution and strong cyclic variation were simultaneously investigated by these three modes.Then, the spray variations under real-engine condition were analyzed. Strong spraycyclic variations were found with the effect of in-cylinder flow. By performing the PODon both the air flow and spray, three types of in-cylinder flow are identified, and resultingin three different kinds of spray pattern. In our previous research within theconstant-volume chamber, three types of fuel spray were identified, namely liquid spray,fully flash-boiling spray and the transitional state. The variations of these three types offuel spray under real-engine condition were quantitatively investigated in this study. Itwas identified that the flash-boiling spray was effective in reducing the cyclic variations.The variation of transitional state spray was significantly strong, which was stronger thanthat of flash-boiling spray by a factor of6.6. The spray variations under different enginespeeds were illustrated as well. So far, the in-cylinder processes before the ignition weresystematically examined.Then, the optical engine was run in firing mode, and the flame kernel formation andthe flame propagation were investigated. Taking flame images within the combustionchamber is a powerful tool to study the combustion quality, particular at the early stageof the combustion. In this study, the in-cylinder pressure and flame image were recordedsimultaneously in high repetition rate (9600Hz) for multiple cycles. The early flameformation variations were illustrated in terms of flame kernel size, location and flamepropagating velocity field. The small difference in early flame formation leads to verydifferent subsequent combustion process and in-cylinder pressure characteristics, whichthe engine power and output are based on. Heat release analysis based upon thein-cylinder pressure, are well correlated with the results from the flame images (Flamearea, POD results of early flame propagating velocity fields). The early flame of lowswirl condition (swirl ratio of0.55) shown stronger variations in terms of flame size,flame kernel position and flame propagating velocity fields. Increasing the swirl ratio to5.68, the variation of flame kernel location was significantly reduced, and the flamekernel was formed two CADs earlier. Moreover, the flame propagating speed increasedby a factor of2, and the variations reduced by half. These lead to the decreased COV(coefficient of variation, nearly50%as well) of IMEP (indicated mean effective pressure)&PP (peak pressure). |
PDF Full Text Request