High-speed imaging analysis of misfires in a spray-guided direct injection engine

https://doi.org/10.1016/j.proci.2010.07.079Get rights and content

Abstract

This study is an experimental investigation of rare misfire and partial burn cycles in a spray-guided spark-ignited direct-injection optical engine. Spark discharge energy, discharge duration, flame imaging, velocity and equivalence ratio were measured every crank angle degree. Imaging was performed using high-speed 2-D particle image velocimetry (PIV) and planar laser inducted fluorescence (PLIF). The engine was operated near its optimum but produced rare and random partial burn and misfire cycles. Spark energy and spark duration for the partial burn and misfire cycles fell within the range of those for the well-burned cycles, with a slight bias toward low-energy, short-duration discharges, indicating no abnormal spark discharge events. PIV and PLIF measurements extracted from a 4 mm × 4 mm region adjacent to and downstream of the spark plug at spark timing revealed that the partial burn and misfire cycles occur under lean mixtures and low velocities, but still within the range of values for the well-burned cycles. PIV and PLIF images of partial burn and misfire cycles were compared to well-burned cycles, which had similar velocities and equivalence ratios near the spark plug at the onset of spark. Observations of the fuel distribution and flame areas for the partial burns and misfires showed that an early flame kernel was always formed, but failed to develop sufficiently to propagate to the fuel in the bowl. A flame kernel arriving late within the piston bowl found significantly leaner conditions and the mixture was not fully consumed leading to a partial burn. For misfire cycles, the mixture in the measurement plane was significantly leaner surrounding the flame kernel, which disappeared shortly after the spark discharge. It is concluded that the partial burns and misfires are not the result of failed ignition, but failure during the flame propagation process.

Introduction

The spray-guided spark-ignited direct-injection (SG-SIDI) engine offers the ability to improve fuel economy by unthrottled load control. The load is lowered by decreasing the fuel (rather than air) to create an overall lean charge, which is locally flammable due to charge stratification. For a SG stratified charge, fuel is injected late in the compression stroke, concentrating an ignitable mixture near the spark plug. Thus, optimum spark timing typically occurs directly after the spray event.

SG-SIDI engines are known to suffer from rare partial burns and misfires, (PB&MFs) [1]. It is observed that the injection event imposes high velocities and large velocity gradients near the spark plug at the time of spark. High velocities can create multiple spark re-strikes, which rapidly changes the location of the spark plasma and shortens the spark duration [1]. Local velocity fluctuations can cause undesirable spark motion and location [2]. The short time duration between the injection and spark event creates highly stratified mixtures including liquid fuel near the spark plug, and non-optimal timing between the injection and spark events are expected to cause an overly rich or overly lean mixture at the spark.

There does not appear to be one universal cause of PB&MFs; instead, they can be caused by multiple parameters [1], [2], [3], [4], [5]. Fuel imaging with planar laser-induced fluorescence (PLIF) has been used in a two-stroke SG-SIDI optical engine to image fuel distribution near the spark plug and have revealed that cyclic variation in fuel concentration near the spark plug are large enough to cause PB&MFs [3]. High-speed fuel LIF imaging was employed to measure the temporal evolution of the fuel cloud leading to misfires [4]. Fuel LIF imaging in a wall-guided direct injection engine showed that non-optimal spray and spark timing can cause over-mixing, producing mixture fractions beyond the flammability limits that lead to misfires [5]. Other LIF studies have shown adequate fuel distribution near the spark plug, but abnormally short spark durations in which a mixture was not ignited [4], [6]. Imaging of the spark and early flame kernel demonstrated that local flow fluctuations can cause unfavorable motion and location, and are a dominant source of random misfire events [2]. Imaging of the spray injection event demonstrated that the resulting high velocities near the spark gap lead to dramatic spark stretching and frequent re-strikes which shorten the spark duration, but did not show any significant correlation to misfire or partial burn cycles [1].

Whereas the previous studies addressed the flow and fuel concentration in separate experiments, this study employs simultaneous fuel and velocity measurements. High-speed particle image velocimetry (PIV) and PLIF of biacetyl tracer are used to measure the flow field and fuel concentrations, respectively, for cases of misfire, partial burn, and normal combustion. Normalized fuel concentration (equivalence ratio) and velocity magnitude data are extracted from a 4 mm × 4 mm region downstream but adjacent to the spark plug at the onset of spark to characterize the initial conditions near the spark plasma. In a separate study on the available spark energy as a function of mixture and flow conditions near the spark plug, relationships were characterized of available spark energy with equivalence ratio, flow velocity, strain rate, and vorticity [7]. Misfiring and partially burning cycles did not exhibit a correlation with either strain rate or vorticity. Therefore, the present study focuses on equivalence ratio and velocity only. The engine was operated to produce rare PB&MFs (five partial burns and nine misfires out of 1392 cycles) so that the other 99.4% of the cycles were near the normal steady-state operating condition. The results demonstrate that the fuel concentration, velocity, spark energy, and spark duration of these rare events were found to fall within the sample population of successful cycles. Thus, as a second step the flow and fuel images of the PB&MFs were compared with well-burned cycles, which have similar velocities and fuel concentrations at the time of spark is performed to further understand the cause of the ignition instabilities.

Section snippets

Experimental

Engine operating conditions, shown in Table 1, were chosen to mimic low-load idle operating conditions in a SG-SIDI engine at 800 rpm. The fuel used in the experiments was a mixture of 90% iso-octane, 10% biacetyl (by volume), which does not alter combustion characteristics in any significant manner [8], [9]. Experiments were conducted without any external dilution (i.e. neither nitrogen nor exhaust gas recirculation (EGR)). The combined high-speed flow and fuel imaging diagnostics have been

Inducement of ignition failures

The goal of this study was to identify in-cylinder events that lead to rare misfires and partial burns at a robust engine idle condition. To find a robust condition, the engine was operated with out external gas dilution of the air, and the end-of-injection (EOI) timing and spark timing were varied to maximize indicated mean effective pressure (IMEP) and minimize the coefficient of variance (COV) of IMEP. Figure 3 which shows the 375 cycle averaged IMEP and COV of IMEP, demonstrates that the

Conclusions

This study of misfires and partial burns in a spray-guided stratified-charge spark-ignition engine provides previously-unavailable, concurrent, cycle-resolved measurements of velocity using 2-D PIV and fuel PLIF during ignition and early flame development. The engine was operated at an idle condition with one degree of spark delay off optimum, which produced rare and random partial burns and misfires (0.3% and 0.5%, respectively); nine misfires and five partial burns occurred within 1392

Acknowledgements

This work was supported by the General Motors Corporation through the General Motors-University of Michigan Collaborative Research Laboratory on Engine Systems Research. The authors are grateful to Arun Solomon, Michael C. Drake and Todd D. Fansler at General Motors and Louise Lu at the University of Michigan.

References (18)

  • M.C. Drake et al.

    Proc. Combust. Inst.

    (1996)
  • V. Sick et al.

    Proc. Combust. Inst.

    (2009)
  • J.D. Smith et al.

    Proc. Combust. Inst.

    (2007)
  • C.M. Fajardo et al.

    Proc. Combust. Inst.

    (2007)
  • M.C. Drake et al.

    Proc. Combust. Inst.

    (2005)
  • T.D. Fansler, M.C. Drake, I. Düwel, F.P. Zimmermann, 7th International Symposium on Internal Combustion Diagnostics,...
  • T.D. Fansler, M.C. Drake, B. Böhm, 8th International Symposium on Internal Combustion Diagnostics, Baden-Baden,...
  • J.D. Smith, V. Sick, SAE Paper 2006-01-1264,...
  • T.F. Alger, J.M. McGee, S.T. Wooldridge, SAE Paper 2004-01-0549,...
There are more references available in the full text version of this article.

Cited by (102)

  • Extension of a thickened flame model to highly stratified combustion—Application to a spark-ignition engine

    2022, Combustion and Flame
    Citation Excerpt :

    G4VDI is an optically accessible experimental engine setup at the University of Michigan, and was designed with the goal of improving understanding of in-cylinder processes in direct-injection spark-ignition engines. This engine has been the focus of recent publications [34–37]. The engine is a four-valve, four-stroke-cycle, single-cylinder, optically accessible spark-ignition engine with direct in-cylinder fuel injection (Fig. 3).

View all citing articles on Scopus
View full text