Figure 5: Identification of
the pre-ignition frequency
through a statistical pro
cess
Table: Main characteristics and settings of cameras ferent combustion indicators as well as
the identification of combustion pat
Camera N°1 Camera N°2 terns that would show a possible deter
Type AVT Pike F100B Phantom Miro 3 minism in pre-ignition phenomenon.
Resolution 1000*1000 800*600
Colour B&W Colour
3 Direct Visualisations
Opening duration (CAD) 9 2
Acquisition per cycle 1 25
3.1 Imaging System
Two CCD cameras were used in this work
to visualise pre-ignition in the combus
tion chamber. Their characteristics are
of this kind of data sample has no choice of the theoretical probability dis summarized in the Table.
chance to be well representative of the tribution (a non exhaustive list of distri The camera was connected to an air-
experimental distribution since it is un butions can be found in [5]). cooled endoscope placed on the flywheel
likely to find a theoretical distribution The comparison of the results ob side of the cylinder head. To protect the
that would take into account both the tained with the first method based on endoscope, a specific sapphire window
asymmetrical aspect due to the slow robust statistics and with the second was inserted into the cylinder head. This
combustions, and the asymmetrical as method based on the iterative modell window has been specifically developed
pect linked to pre-ignitions. ings has always yield satisfactory results for our application and allows the visuali
The quantification of the pre-ignition so far. These tools also have the basic ad sations of a violent combustion ignited by
frequency and the correct modelling of vantage to consider that the underlying a pre-ignition. The field of view obtained
normal combustions ask thus to set aside distribution is not symmetric. This main by our optical system is given in Figure 6.
abnormal combustions. This sorting step feature explains why they lead to an ac One objective of this direct visualisa
can be obtained through an iterative and curate quantification of pre-ignition and tion is to determine the spatial origin of
automatic process consisting in perform justifies their use when different engines the pre-ignition. This information can
ing successive statistical fittings remov or impacts of different settings on the help us discuss the potential causes of
ing at each step of this iterative process same engine are to be compared. That is abnormal combustions. All visualisa
the cycle which has the lowest 10 % MFB also why the first tool based on robust tions and results presented here were
angle and which is then potentially a statistics has been associated to direct obtained with a gasoline single cylinder
pre-ignition. visualisations to extract more efficiently engine operated at a speed of 1000 rpm
The judicious choice of the theoreti the relevant data regarding pre-ignition. at full load.
cal probability distribution then allows Additionally, some other tools have also
to efficiently define the limit between been developed at the same time to com 3.2 Methodology of Extraction
normal combustions and pre-ignitions. plete our analysis toolbox. Various meth A pre-ignition phenomenon is a sporadic
Two approaches are conceivable either odologies have been defined concerning event and the beginning of this abnor
by defining a quality index of the succes the exploitation of the links between dif mal combustion can occur in a cycle
sive modellings or by tracking the evolu
tions of the parameters defining the cho
sen probability distribution. In the first
case, the pre-ignition frequency is deter Figure 6: Field
mined by the numbers of cycles that of view in the
should be removed from the original da combustion
ta sample to reach the maximal relative chamber
quality index, Figure 5, lower left-hand
corner). In the second case, the pre-igni
tion frequency can be determined thanks
to particular values of the parameters de
fining the chosen probability distribu
tion. The progressive removal of cycles
having the lowest 10 % MFB angles in
deed unveils some particular values like
maximal values or inflexion points only
when all the pre-ignitions have been re
moved, Figure 5, lower right-hand cor
ner. The particular value that has to be
analysed depends on the choice of the
parameters and also of course on the
Figure 8: Examples
of pre-ignition with
different locations
zone behind the spark plug plan. Acqui and to follow the whole combustion pre-ignition phenomenon and will be
sition of images with another field of process. Using this kind of camera im probably be studied in our future work.
view could usefully complement the proves the productivity of results be
map of pre-ignition zone. cause there is at least one usable image
for each pre-ignition cycle, Figure 7/step 6 Outlook
2. We present on the Figure 10 the mean
5 High Speed Camera Potential rate of heat release ROHR on 300 cycles Several hypotheses have already been for
and the ROHR of a pre-ignition cycle (cy mulated to explain the potential causes
Camera N°1 has been used with a long cle N°250). This pre-ignition cycle is very of pre-ignition [7]. Nevertheless, the in
exposure time (9 CAD) for two main rea fast and violent but with the camera teraction between these different possi
sons: N°2, we can split the ROHR and obtain a ble causes makes the analysis and the
– to obtain enough signal (depends on lot of interesting images. Despite the control really tough. Resultingly, pre-
the camera sensitivity) small exposure time (2 CAD), the signal ignition remains a critical issue during
– to increase the potential to capture an is sufficient and we can precisely analyse the development on new highly boosted
abnormal combustion. the spatial origin and the propagation spark ignition engines.
The net advantage of an high speed cam of this abnormal combustion. Two innovative statistical approaches
era resides in gathering a lot of images Another information given by the cam have been used to develop reliable in
in one cycle. It thus becomes easier to era N°2 is colour. Its analysis could be used dexes and methodologies, allowing a
capture the beginning of a pre-ignition to go further in the understanding of the precise quantification of the pre-igni
Figure 9: Pre-ignition
zone – case a: sample
of 101 images and
case b: sample of 132
images
References
[1] Vangraefschepe, F.; Zaccardi, J.-M.: “Analysis of
destructive abnormal combustions appearing at
high load and low engine speed on high perfor
mance gasoline engine“, The Spark Ignition Engine
of the Future, SIA Congress, 2007
[2] Zaccardi, J.-M.; Duval, L.; Pagot, A.: “Development
of Specific Tools for Analysis and Quantification of
Pre-ignition in a Boosted SI Engine“, SAE Paper
2009-01-1795
[3] Manz, P.-W.; Daniel, M.; Jippa, K.-N.; Willand, J.:
“Pre-ignition in highly-charged turbo-charged
engines. Analysis procedure and results“, 8. Inter-
nationales Symposium für Verbrennungsdiagnostik,
Figure 10: Continuous record of a pre-ignition event Baden-Baden, 2008
[4] Huber, P. J.: “Robust Statistics“, John Wiley and
Sons, New-York, 1981
[5] Saporta G. : “Probabilités, analyse des données et
statistique“, Technip, 2006
tion frequency. It is now possible to We prove the deep interest in combin [6] Chaux, C. ; Duval, L.; Benazza, A.; Benyahia, J.;
compare different configurations and ing statistical analysis and direct visuali Pesquet, J.: “A nonlinear Stein based estimator for
to guide the design of new engines on sation. Thanks to our experimental set multichannel image denoising“, IEEE Transactions
on Signal Processing 56, Nr. 8, S. 3855-3870, 2008
test bench. It is also conceivable to trans up and methodology, we make up a map
[7] Willand, J.; Daniel, M.; Montefrancesco, E.;
pose some of these tools to online analy of preferential zone of pre-ignition and Geringer, B.; Hofmann, P.; Kieberger, M.: „Grenzen
ses either for steady state operations or we show that these zones depend on the des Downsizing bei Ottomotoren durch Vorentflam-
for transients. experimental conditions. Such maps mungen“, MTZ Mai 2009