Multiple Pulse Laser Ignition in GDI Lean Combustionpckuang/Publications/Zheng Kuang's... ·...
Transcript of Multiple Pulse Laser Ignition in GDI Lean Combustionpckuang/Publications/Zheng Kuang's... ·...
-
1
Multiple Pulse Laser Ignition in GDI Lean Combustion
Hua Cheng1)
, Zheng Kuang2)
, Vincent Page1)
, Elliott Lyon2)
, Geoff Dearden2)
, Tom Shenton1)
1). Powertrain Control Group, Centre for Engineering Dynamics, University of Liverpool, Liverpool L69 3GH, UK
2). Laser Group, Centre for Material and Structures, University of Liverpool, Liverpool L69 3GQ, UK
Abstract: Laser Ignition (LI) is a new, innovative technology which overcomes several
limitations of conventional spark ignition; it presents many potential advantages for engine
dynamic control, especially for Gasoline Direct Injection (GDI) Engine. This paper highlights
a new approach, using Principle Component Analysis (PCA) of the Pressure Curve, to
dynamically control Multiple Pulse Laser Ignition (MPLI); and reports on the effectiveness of
Multiple Pulse Laser Ignition (MPLI) in lean combustion. Experimental test data is presented;
and these results indicate how the control of MPLI improves the performance of the engine.
This work is part of an EPSRC funded research project which is to investigate how optimized
control strategies, including multiple stratified fuel injection (MSFI); MPLI and Multiple
Positions Laser Ignition (MPsLI), can help to improve the stability of leaner operation and
reduce the emissions. The possible future of LI dynamical control strategies (DCS) is
discussed.
1. Introduction
Automotive manufacturers are dedicated to improving engine performance, while maintaining fuel economy and
reducing exhaust emission. In recent decades, due to the energy supply limitations and environmental damage
concerns, the GDI engine has been introduced to replace the port fuel-injection (PFI) engine [1]-[2]. However, in
previous research, it has been found that GDI has some significant limitations, which require more complicated
and intelligent DCS to improve the stability and performance to achieve its basic objective of saving fuel and
reducing emissions.
The conventional spark plug has fixed ignition locations, limiting the spatial variations and requires long
discharge dwell time (ms), making DCS of Spark Ignition (SI) engines with multiple ignitions at different times
difficult. With these limitations the manufacturers are therefore aiming to improve the quality of atomization of
fuel spray [3]-[5] or to develop new types of engine by using Homogeneous Charge Compression Ignition
(HCCI) [6]-[8] which removes the spark plug. The LI engine system provides an opportunity for DCS to control
the ignition, either temporally or spatially thereby lifting many of the restrictions of the spark plug and opening
up possible improvements to engine performance and efficiency [9]-[10].
Since the first reporting of LI tests on an ICE (Internal Combustion Engine) by Dale et al in 1978 [11],
engine based LI research has been primarily led by the groups of McIntyre and Woodruff, Wintner [12], and
Dearden and Shenton [13]-[15]. The LI system has been studied and investigated by the Laser Group and
Automotive Powertrain Control Group of the University of Liverpool (UoL) for many years. Current LI systems
have been stably applied and tested on both a research laboratory prototype engine and a production engine. In
practice, the stability of the LI engine rivals that of traditional SI engines; the CoV (Coefficient of Variation) of
IMEP (Indicated Mean Effective Pressure) of both engines under optimal conditions are less than 2%.
Additionally the LI engine is more stable in lean combustion than SI engine.
In this paper, the goal of multiple pulse control is to allow the second pulse to be fired at the appropriate time
to improve the performance and stability. With reference to Fig. 1, if the first pulse is fired and good combustion
occurs, then the curve in green will result. However, if the first pulse is fired and poor combustion occurs then
the red dotted pressure curve will result. If the cylinder pressure of the engine is monitored and compared to a
desired pressure curve, a decision can then be made as to whether, and possibly when, the second pulse is fired.
If the second pulse is fired after unsuccessful combustion then the blue pressure curve will result.
A critical question arises at this point. Why not always fire the second pulse? If the second pulse is fired,
then the purple dotted lines in Fig. 1 results. This is because the second pulse will accelerate the combustion too
much and create turbulent combustion and knocking. This reduces efficiency and combustion performance. This
affect will be discussed later in the paper.
-
2
Fig. 1. Pressure curves for combustion with and without secondary pulse
2. Experimental LI engine system
The Automotive Powertrain Control Group of the UoL has a strong background and experience in calibration
and optimization of DCS for IC engines. The Uol team has a long relationship in collaboration with Ford Motor
Co. (FMC) and Jaguar Land Rover (JLR). The first successful application of LI to an ICE at UoL was on a 4
cylinder Ford Zetec petrol engine in 2006. For research purposes, a new prototype GDI Single Cylinder Engine
(SCE) has been supported and built by Ford US in 2009. In this paper, tests were carried out using the SCE,
which has a 776 cc swept volume and a compression ratio of 9.2. The dSpace Hardware in Loop (HIL) system
with fast DSP-slave (Digital Signal Processor) has been used as a rapid prototyping system to achieve control
and collect real time data from the engine. In the system, a BOSH ES-HDEV-1 fuel injection driver ECU with a
100 bar pressure fuel pump is applied to generate high quality in-cylinder fuel sprays (Fig. 2). An AVL M5
pressure sensor and Kistler 5011 type charge amplifier are used to enable sensitive in-cylinder pressure data
measurement. The experimental setup is shown schematically in Fig. 2.
Fig. 2. SCE prototyping engine for LI research (The collimated laser beam is guided into the cylinder head using reflective mirror, the
LI plug is used to replace the SI plug and create the ignition)
-
3
For the LI, a Litron LPY-764-30 laser and a Spectron Mini-Q laser (both Q-switched Nd:YAG) are used to
create the first and second pulses respectively. The laser wavelength can be switched between 532 nm or 1064
nm (with or without 2nd
harmonic generators). To achieve the MPLI, two types of beam splitter have been used
to obtain multiple beams: one beam splitter is a polarizing cube to couple the same wavelength beam (Fig. 3a);
another is the wavelength beam splitter which allows the combining of different wavelength beams (Fig. 3b).
( a )
( b )
Fig. 3. MPLI optical setup diagram (Note: the Polarizing beam splitter (a) can be replaced by the Wavelength beam splitter (b) to couple
different wavelength beams of 532 nm and 1064 nm)
The multiple beams are aligned to the same optical path through the beam splitter. Then the laser is guided to
the engine cylinder by reflective mirrors. An optical LI plug from a previous research project uses a sapphire
front window to protect the focusing lens. During the work at Liverpool using this plug system, cracks and soot
deposits (Fig. 4a left) appeared on its surface due to the combustion. This therefore created the need for a
self-cleaning pulse to clear the surface windows to ensure the laser beam quality.
(a) (b)
Fig. 4. (a) Comparison of effects in use between protection windows and lens. (b) The second generation of optical plug with only lens but
no front windows
-
4
In new research, it has been found that the front windows can be removed and then a 3 mm thick sapphire
lens will not take damage even over long periods of operation. In this new design, the lens is then still very clean
and with no damage at all after the engine running over 20 hours (Fig. 4a right). The second generation (Fig. 4b)
design of the plug has overcome the previous crack problems and does not need the self-cleaning pulse anymore.
3. In-cylinder Combustion Feedback Control (ICCFC) methodology
PCA is a statistical approach that uses orthogonal transformations to convert a set of observations of possibly
correlated variables into a set of values of linearly uncorrelated variables called principal components. It has
been used to analyze and decompose large and complicated profiles of signals [16]-[17].
Fig. 5. The PCA for Pressure decomponent and combustion quality identify
The use of PCA can reduce the signal component dimensions and group the signals, for instance arising from
pressure variance (Fig. 5 a&b). In fact, the PCA can help to identify the quality of combustion through the
pressure signal; misfires, poor combustion and good combustion can be easily recognized (Fig. 5c).
During the experiment, it was discovered that always firing both lasers will sometimes cause increased
turbulent flame propagation and initiate knocking. Therefore, to avoid knocking and achieve the objective of real
time identification of combustion quality, the STD (standard deviation) function (eq. 1 and eq. 2) of Matlab from
the PCA output is used in the control map for the second pulse control to trigger at the optimized time. From the
investigation, it was found that the minimum dwell timing can be less than 6˚CAs (Crank Angles). But it is
suggested that 10˚ is used as it depends on the first LI trigger CAs (Fig. 6b).
̅ ∑
(1)
(b) (a)
(c)
-
5
√∑ ( ̅)
(2)
Neural-Networks (NN) is a popular tool for use in feedback control for complex non-linear mapping in
optimal control strategies. In our research, a novel PCA-NN approach is used to find optimized maps for MPLI
DCS.
(a) (b)
Fig. 6. (a) Pressure variations between reference, very poor, poor and good combustion. (b) STD values bewteen different pressure curves.
4. Results
During the experiment, a large volume of data was collected with the engine operating at varying speeds, loads
and air-fuel ratios. Part of the results for a homogenous charge operation has been shown in this paper. Fig. 7
shows curves of in-cylinder pressure at lambda =1.3, the engine is set at 1500 rpm, Break Mean Effective
Pressure (BMEP) = 4.8 bar and with Laser energy = 30 mJ only. After the investigation, two points of interest
are noted. First, the misfire has been mostly removed by inducing the second pulse. Second, the poor combustion
has been partially improved (Fig. 7). Fig. 8 shows a and clearly presents the improvement of engine
stability while using dual pulse LI. The has been reduced to about 5% from 10%.
Lambda = 1.3
(a) Pressure curve with only a single ignition for comparing with dual pulse pressure results below (lambda = 1.3)
0 100 200 300 400 500 600 700
0
2
4
6
8
10
12
14
16
18
20
22
Pressure curve with a single pulse & triggered at 22 BTDC
Crank Angle
Pre
ssure
(B
ar)
-
6
(b) Pressure curve with a dual pulse and second pulse triggered at different BTDC angles
Fig. 7. The pressure varying curves at lambda=1.3. (Note: figure (a) is only a single laser ignition as a reference for comparing with dual
pulse; figure (b) present dual pulse ignition and the second ignition triggered at different BTDC crank angle.)
Fig. 8. The stability variations (Note: the first point at 22˚ is only single igntion)
Fig. 9 below demonstrates the pressure curve variations of dual pulse laser ignition and the improvement of
combustion quality at lambda=1.5. The engine is running at an average speed of 700 rpm but with the same
settings as lambda=1.3. It also presents improvement of combustion quality. However, due to the very lean
fuel-air ratio, the dual pulse has slightly less power output than lambda 1.3. The Fig. 10 shows the CoV changes
0 100 200 300 400 500 600 700
0
2
4
6
8
10
12
14
16
18
20
22Dual pluse at lambda 1.3 & second pulse triggered at 21 BTDC
Crank Angle
Pre
ssure
(B
ar)
0 100 200 300 400 500 600 700
0
2
4
6
8
10
12
14
16
18
20
22Dual pluse at lambda 1.3 & second pulse triggered at 19 BTDC
Crank Angle
Pre
ssure
(B
ar)
0 100 200 300 400 500 600 700
0
2
4
6
8
10
12
14
16
18
20
22Dual pluse at lambda 1.3 & second pulse triggered at 17 BTDC
Crank Angle
Pre
ssure
(B
ar)
0 100 200 300 400 500 600 700
0
2
4
6
8
10
12
14
16
18
20
22Dual pluse at lambda 1.3 & second pulse triggered at 15 BTDC
Crank Angle
Pre
ssure
(B
ar)
0 100 200 300 400 500 600 700
0
2
4
6
8
10
12
14
16
18
20
22Dual pluse at lambda 1.3 & second pulse triggered at 12 BTDC
Crank Angle
Pre
ssure
(B
ar)
0 100 200 300 400 500 600 700
0
2
4
6
8
10
12
14
16
18
20
22Dual pluse at lambda 1.3 & second pulse triggered at 9 BTDC
Crank Angle
Pre
ssure
(B
ar)
0 100 200 300 400 500 600 700
0
2
4
6
8
10
12
14
16
18
20
22Dual pluse at lambda 1.3 & second pulse triggered at 7 BTDC
Crank Angle
Pre
ssure
(B
ar)
0 100 200 300 400 500 600 700
0
2
4
6
8
10
12
14
16
18
20
22Dual pluse at lambda 1.3 & second pulse triggered at 2 BTDC
Crank Angle
Pre
ssure
(B
ar)
0 100 200 300 400 500 600 700
0
2
4
6
8
10
12
14
16
18
20
22Dual pluse at lambda 1.3 & second pulse triggered at 0 BTDC
Crank Angle
Pre
ssure
(B
ar)
22 21 19 17 15 12 9 7 2 0 4
5
6
7
8
9
10
11
10.5554
9.7549
6.031
4.90944.6244
5.3238
5.7192
4.9635.3371
5.1792
COV__IMEP variables of dual pulse at lambda = 1.3
Second pulses trigger timing (BTDC Crank Angle)
Cov__S
tabili
ty (
%)
-
7
by dual pulse LI. It has been improved from 20.7% at 22 to 10.1% at 2 . Meanwhile, the results also indicate the control of second pulse trigger and subsequent selection of secondary pulse timing will benefit the
combustion quality.
Lambda = 1.5
(a) Pressure curve with only a single ignition for comparing with dual pulse pressure results below (lambda = 1.5)
Fig. 9. The pressure varying curves at lambda=1.5. (Note: figure (a) is only a single laser ignition as a reference for comparing with dual
pulse; figure (b) present dual pulse ignition and the second ignition triggered at different BTDC crank angle.)
0 100 200 300 400 500 600 700
0
2
4
6
8
10
12
14
16
18
20
22
24
Pressure curve with a single pulse & triggered at 22 BTDC
Crank Angle
Pre
ssure
(B
ar)
0 100 200 300 400 500 600 700
0
2
4
6
8
10
12
14
16
18
20
22
24
Dual pluse at lambda 1.5 & second pulse triggered at 21 BTDC
Crank Angle
Pre
ssure
(B
ar)
0 100 200 300 400 500 600 700
0
2
4
6
8
10
12
14
16
18
20
22
24
Dual pluse at lambda 1.5 & second pulse triggered at 19 BTDC
Crank Angle
Pre
ssure
(B
ar)
0 100 200 300 400 500 600 700
0
2
4
6
8
10
12
14
16
18
20
22
24
Dual pluse at lambda 1.5 & second pulse triggered at 17 BTDC
Crank Angle
Pre
ssure
(B
ar)
0 100 200 300 400 500 600 700
0
2
4
6
8
10
12
14
16
18
20
22
24
Dual pluse at lambda 1.5 & second pulse triggered at 15 BTDC
Crank Angle
Pre
ssure
(B
ar)
0 100 200 300 400 500 600 700
0
2
4
6
8
10
12
14
16
18
20
22
24
Dual pluse at lambda 1.5 & second pulse triggered at 12 BTDC
Crank Angle
Pre
ssure
(B
ar)
0 100 200 300 400 500 600 700
0
2
4
6
8
10
12
14
16
18
20
22
24
Dual pluse at lambda 1.5 & second pulse triggered at 9 BTDC
Crank Angle
Pre
ssure
(B
ar)
0 100 200 300 400 500 600 700
0
2
4
6
8
10
12
14
16
18
20
22
24
Dual pluse at lambda 1.5 & second pulse triggered at 7 BTDC
Crank Angle
Pre
ssure
(B
ar)
0 100 200 300 400 500 600 700
0
2
4
6
8
10
12
14
16
18
20
22
24
Dual pluse at lambda 1.5 & second pulse triggered at 2 BTDC
Crank Angle
Pre
ssure
(B
ar)
0 100 200 300 400 500 600 700
0
2
4
6
8
10
12
14
16
18
20
22
24
Dual pluse at lambda 1.5 & second pulse triggered at 0 BTDC
Crank Angle
Pre
ssure
(B
ar)
-
8
Fig. 10. The stability variations (Note: the first point at 22˚ is only single ignition.)
In previous tests, it was found that the minimum break down energy to create a spark with the laser was 4 mJ
in-cylinder [15]. The air-break down energy is stable at 15 mJ but with weak sparking when used as an ignition
source. The 15 mJ level however can run the engine in stable operation. The 30 mJ level is very stable in
standard engine air-fuel ratios.
5. Conclusion
A new EPSRC project named In-Combustion-Event Feedback (ICEF) control by LI of UoL team has been
recently launched. The objective is to design and map the DCS to globally control the LI ignition in both
temporal and spatial modes. The experimental results from this paper were obtained as part of this project. These
strongly indicate that the MPLI will benefit the combustion quality in lean conditions. The CoV stability of the
engine performance can be improved in dual pulse LI by up to 30% (not including the misfires).
6. Reference
[1] F. Zhao, D.L. Harrington, M.-C. Lai, Automotive Gasoline Direct-Injection Engines, (SAE international, Warrendale, Pa.), chapter 1, 1999.
[2] F. Zhao, M.-C. Lai, and D.L. Harrington, “Automotive spark-ignited direct-injection gasoline engines,” Progress in Energy and
Combustion Science, 25, 437-562 (1999). [3] J. Lee, M. Yamakawa, S. Isshiki, and K. Nishida, “An Analysis of Droplets and Ambient Air Internation in a D.I. Gasoline Spray Using
LIF-PIV Technique,” SAE Technical Paper, 02-24, (2002).
[4] G. Bianchi, C. Forte, S. Negro, and P. Pelloni, “A 1d Model of the Prediction of Flash Atomization in Gdi Multi-Hole Injectors: Preliminary Results,” SAE Int. J. Engines 1 (1), 1278-1293 (2009).
[5] H. Husted, T. Spegar, and J. Spakowski, “The Effects of GDi Fuel Pressure on Fuel Economy,” SAE Technical Paper, 2014-01-1438,
(2014). [6] S. Tanaka, A. F. Ayala, J. C. Keck, and J.B. Heywood. "Two-stage ignition in HCCI combustion and HCCI control by fuels and
additives." Combustion and flame 132, no. 1, 219-239 (2003).
[7] S. Tanaka, A. F. Ayala, and J. C. Keck, "A reduced chemical kinetic model for HCCI combustion of primary reference fuels in a rapid compression machine." Combustion and flame 133, 467-481, (2003).
[8] H. Machrafi, and S. Cavadias, "Three-stage autoignition of gasoline in an HCCI engine: An experimental and chemical kinetic modeling
investigation." Combustion and flame 155(4), 557-570, (2008). [9] T.X. Phuoc, “Laser-induced spark ignition fundamental and applications.” Optics and Lasers in Engineering, 44(5), 351-397 (2006).
[10] M.H. Morsy, “Review and recent developments of laser ignition for internal combustion engines applications.” Renewable and
Sustainable Energy Reviews, 16(7), 4849-4875 (2012). [11] J.D. Dale, M.D. Checkel, and P.R. Smy, “Application of high energy ignition systems to engines.” Progress in energy and combustion
science, 23(5), 379-398 (1997).
[12] D.L. McIntyre, S.D. Woodruff, and J.S. Ontko, “Lean-burn stationary natural gas engine operation with a prototype laser spark plug.” Journal of Engineering for gas turbines and power, 132(7), 072804 (2010).
[13] G. Dearden, and A.T. Shenton, “Laser ignited engines: progress, challenges and prospects.” Optics Express (Energy Express), 21-S6,
A1113-A1125 (2013). [14] S.D. Carroll, R.D. Dodd, A.T. Shenton, G. Dearden, S. Scarisbrick and S. Keen, “Laser Ignition in an Automotive Engine: Analysis and
Control of C.O.V. in I.M.E.P. and P.P.P..” Proceedings of Japanese SAE Annual Congress 2005, Japanese Society of Automotive
Engineering (SAE), 54190-54201 (2005). [15] P.B Dickinson, A.T. Shenton, J. Mullett, G. Dearden and A. Scarisbrick, “Prospects for Laser Ignition in Gasoline Control.” 10th
International Symposium on Advanced Vehicle Control, 1-6 (2010).
[16] H. Hotelling, “Analysis of a complex of statistical variables into principal components.” Journal of Educational Psychology, 24, 417-441, and 498-520 (1933).
[17] I.T. Jolliffe, Principal Component Analysis, second edition Springer-Verlag New York, 2002.
22 21 19 17 15 12 9 7 2 0 10
12
14
16
18
20
22
20.6768
14.1531
17.6202
12.2373
16.6087
13.8996
15.4699
11.7051
10.1988
15.0483
COV__IMEP variables of dual pulse at lambda = 1.5
Second pulses trigger timing (BTDC Crank Angle)
Cov__S
tabili
ty (
%)
http://en.wikipedia.org/wiki/Journal_of_Educational_Psychology