Control of laser ignition in an internal combustion engine

IOP PUBLISHING

JOURNAL OF PHYSICS D: APPLIED PHYSICS

J. Phys. D: Appl. Phys. 40 (2007) 4730–4739

doi:10.1088/0022-3727/40/15/056

The influence of beam energy, mode and focal length on the control of laser ignition in an internal combustion engine J D Mullett1 , R Dodd1 , C J Williams1 , G Triantos2 , G Dearden1 , A T Shenton2 , K G Watkins1 , S D Carroll3 , A D Scarisbrick3 and S Keen4 1 Laser Group, Department of Engineering, University of Liverpool, Brownlow Street, Liverpool, L69 3GH, UK 2 Powertrain Control Group, Department of Engineering, University of Liverpool, Brownlow Street, Liverpool, L69 3GH, UK 3 Ford Motor Company, Dunton Research and Engineering Centre, Laindon, Basildon, Essex, SS15 6EE, UK 4 GSI Group, Cosford Lane, Swift Valley, Rugby, Warwickshire, CV21 1QN, UK

E-mail: [email protected]

Received 20 February 2007, in final form 7 June 2007 Published 20 July 2007 Online at stacks.iop.org/JPhysD/40/4730 Abstract This work involves a study on laser ignition (LI) in an internal combustion (IC) engine and investigates the effects on control of engine combustion performance and stability of varying specific laser parameters (beam energy, beam quality, minimum beam waist size, focal point volume and focal length). A Q-switched Nd : YAG laser operating at the fundamental wavelength 1064 nm was successfully used to ignite homogeneous stoichiometric gasoline and air mixtures in one cylinder of a 1.6 litre IC test engine, where the remaining three cylinders used conventional electrical spark ignition (SI). A direct comparison between LI and conventional SI is presented in terms of changes in coefficient of variability in indicated mean effective pressure (COVIMEP ) and the variance in the peak cylinder pressure position (Var PPP ). The laser was individually operated in three different modes by changing the diameter of the cavity aperture, where the results show that for specific parameters, LI performed better than SI in terms of combustion performance and stability. Minimum ignition energies for misfire free combustion ranging from 4 to 28 mJ were obtained for various optical and laser configurations and were compared with the equivalent minimum optical breakdown energies in air. (Some figures in this article are in colour only in the electronic version)

1. Introduction Recent research in laser-induced ignition (LI) of air–fuel mixtures in internal combustion (IC) engines has shown there to be many potential advantages over conventional electrical spark ignition (SI) [1–4]. Spark plugs offer only limited possibilities for optimizing engine efficiency, due to their fixed position within a cylinder and the protrusion of electrodes which disturb the cylinder geometry and can quench 0022-3727/07/154730+10$30.00

© 2007 IOP Publishing Ltd

the flame kernel. Laser radiation is non-invasive and has greater flexibility in terms of the ignition position, allowing the possibility of multi-point ignition [5]. Other potential benefits of LI include reduced emissions, faster ignition, more stable combustion, lower idle speeds and better cold engine performance, when compared to conventional SI. A review of the literature has shown there to be four principal LI mechanisms: non-resonant breakdown ignition, resonant breakdown ignition, thermal ignition and

Printed in the UK

4730

Control of laser ignition in an internal combustion engine

photochemical ignition [1, 3]. Non-resonant breakdown is the mechanism by which LI is performed in the tests presented in this paper and is the most widely used and studied form of LI. The electrical field of a focused laser beam is sufficient to cause dielectric breakdown of the air–fuel mixture, using laser irradiances in excess of 1011 W cm−2 . The process generally begins with multi-photon ionization of a few molecules, which leads to the release of electrons that can then readily absorb more photons from the laser source via the inverse bremsstrahlung process. This in turn increases the kinetic energy of these released electrons, which then collide with other molecules and ionize them, leading to an electron avalanche and subsequent breakdown of the combustible gas mixture. Non-resonant breakdown LI is similar to conventional electric SI, in that a plasma is produced which emits light, heat and a shockwave, although laserinduced sparks are smaller in size, shorter in duration and have higher temperatures and densities. The large majority of previous studies on LI have investigated the fundamental processes of laser-induced gas breakdown for the application of gas reciprocating engines, where mixtures of methane, hydrogen and air are most commonly used. However, relatively few studies have concentrated on LI in automotive gasoline IC engines [2, 6–9], which is the main focus of this paper. Research conducted at The University of Liverpool [10–12], is, to the authors’ knowledge, the only LI research reported to date to use an otherwise unmodified production automotive engine. Moreover, previous LI studies have used relatively long focal length (FL) lenses to focus the beam through a port window of a combustion chamber, with FLs ranging from 50 to 450 mm [2, 4, 6, 13]. The beam energy required to create a plasma is higher for longer FL lenses for a given beam diameter, due to the larger minimum waists produced. Furthermore, the specific location of the plasma varies to a greater degree along the path of the laser beam as the focal point volume is increased. This study therefore investigates LI using shorter FL lenses ranging from 15 to 36 mm, which allows the optical plug to be compact in design, as the tight focuses achieved means that beam expansion is not required. In addition, the effect of beam quality on LI has been overlooked in recent research, as highlighted by Phuoc [3]; consequently this area has also been addressed in this paper. The main aims of the research discussed in this paper are to investigate the effects of laser parameters, specifically beam energy, beam quality and FL, on the control of LI and to assess the performance of LI against conventional SI. This is achieved by examining the engine’s combustion stability in terms of cycle to cycle variation (CCV), in two cylinders (one ignited by a conventional spark plug and the other by an Nd : YAG laser beam). CCV occurs due to variation in the air–fuel ratio near the ignition point, ignition timing, and formation of the flame kernel at combustion initiation. Benefits of LI associated with engine performance are most likely to be a result of reduced CCV. The indicators of CCV used in this work were indicated mean effective pressure (IMEP) and peak pressure position (PPP), which have the units of bar and degrees respectively. IMEP is a fictitious average pressure exerted on the piston during the expansion stroke of a cycle, and is defined as IMEP =

Wi , Vd

where Vd is the displaced cylinder volume and Wi is the gross work delivered to the piston over the compression and expansion strokes, which is obtained by the circular integration of the pressures, P , over these strokes, with respect to cylinder volume [14]: Wi = P dV . IMEP is extensively used in engine calibration and the coefficient of variation in IMEP (COVIMEP ) is used as an indicator of combustion stability. The COVIMEP is commonly used in industry and is defined as σIMEP × 100, COVIMEP = IMEP where σIMEP is the standard deviation in IMEP and IMEP is the mean IMEP. The COVIMEP is given as a percentage and defines the variability in indicated work per cycle. Vehicle driveability problems usually result when the COVIMEP exceeds about 10% [15]. The PPP is the crank angle in degrees after top dead centre (ATDC), at which the peak pressure occurs for each combustion event. The optimum PPP is usually around 16◦ ATDC, but is mainly dependent on engine geometry. The variance in PPP (Var PPP ) has also been demonstrated as an indicator of CCV [16] and is defined as Var PPP = (σPPP )2 , where σPPP is the standard deviation in PPP. Lower values of COVIMEP and Var PPP indicate reduced CCV, and hence better combustion performance and stability [15].

2. Experimental The laser used was a ‘Mini-Q’ flashlamp pumped Q-switched Nd : YAG, manufactured by GSI Group, operating at the fundamental wavelength of 1064 nm. The laser was operated in three sets of offline and online experiments, where each set was performed using a different laser cavity aperture with the diameters of 1.3, 2 and 3 mm. Using larger cavity apertures allowed greater beam energies to be tested while compromising on beam quality and focusablity. The beam pulse width was measured using an Agilent 54641A oscilloscope and an Alphalas UPD-300IR1 photodiode, and was found to be 10 ns to the full width half maximum for each cavity aperture. A schematic of the experimental setup is illustrated in figure 1, which shows two optical legs to the optical system arrangement, one for offline testing and the other for online testing. The laser was operated at a pulse repetition rate of 12.5 Hz for both the offline and online experiments, which corresponds to an engine speed of 1500 rpm.

Figure 1. Experimental setup for offline and online testing of the prototype LI system.

4731

J D Mullett et al (a)

Figure 2. Photograph of a disassembled optical plug.

2.1. Offline optical tests Five different FL lenses (15, 18, 24, 30 and 36 mm) were tested individually in a specifically designed optical plug. These were all 6 mm diameter, uncoated BK7 plano-convex lenses, apart from the 36 mm FL lens, which had a visible to near infrared coating. A clean uncoated sapphire window of 1 mm thickness and 5 mm diameter was sealed at the bottom of the optical plug for each of the different lenses. A photograph of the optical plug can be seen in figure 2 showing all the components in order of assembly along the central axis. The minimum beam waist produced by each lens was positioned at 4 mm from the bottom of the plug (which is at the same location as the electrical discharge of the spark plugs), as this was found from previous testing to be the optimum LI position for this engine. Mirror (1) was installed on the optical bench for offline testing to direct the beam into the optical plug, as shown in figure 1. The plug was at a beam path length of 1.4 m from the laser head which was the same distance as it was for the online tests. An energy meter (Gentec ED 200 head and Solo PE monitor) was used to measure the laser pulse energies at various laser drive levels, taking the average energy of a 200 pulse sample. The energies were measured before each optical plug on the bench, starting with a 40% laser drive level, and then increasing by set increments up to 100%. The subsequent energies transmitted through the plugs for each energy increment were also measured by placing the energy meter at 40 mm after the plug, accounting for the plasma produced. From these tests, the transmissions of the optical plugs were measured in order to calculate the actual beam energy that would be delivered inside the engine’s cylinder. An Electrophysics ‘Micro-viewer’ 7290A charged coupled device camera system was used to measure the minimum spot sizes produced by the different FL lenses in the optical plug to the 1/e2 limit to an accuracy of ±5 µm. These minimum spot sizes were used with the recorded energies to calculate laser irradiances in GW cm−2 for the various settings. The camera was also used to measure the beam sizes and capture the beam profiles on the lenses in the optical plug for laser drive levels between 40% and 100%, to ensure that these were similar, as a variance in beam size would affect the minimum spots produced. The beam quality factor M 2 was calculated for each of the laser cavity apertures with these measured beam sizes, using the formula M2 =

dπ D , 4f λ

where d is the minimum beam diameter produced by the focusing lens, D is the beam diameter incident on the focusing lens, f is the FL of the lens and λ is the wavelength. 2.2. Online optical tests For the online testing, the laser system was controlled through a dSpace DS1005 card in a bus linked expansion box, using 4732

(b)

(c)

Figure 3. Cross-sectional beam intensity profiles at 1.4 m from the laser output, produced using laser cavity apertures of diameter: (a) 1.3 mm, (b) 2 mm and (c) 3 mm.

an open loop Simulink laser timing model designed and run through MATLAB. The IC engine used was an unmodified four-stroke 1.6 litre Zetec test engine from a Ford ‘Mondeo’. This had 4 cylinders, 16 valves and double overhead camshafts, with aspirated port fuel injection (PFI) and operated in a homogeneous ignition mode. The engine was connected to a low inertia dynamometer, to provide a load representative of in-service conditions. Cylinder pressure data was taken from cylinders 1 and 4, using Kistler engine pressure sensors in situ, for comparison of LI to SI combustion cycles. During testing, cylinder 1 was fired optically using the laser, while cylinders 2, 3 and 4 were ignited using conventional spark plugs, which were laser platinum premium manufactured by NGK. For each cycle, the Simulink model sent a signal via dSpace to the laser to activate the flashlamp, where 100 µs later the Q-switch was triggered internally by the laser power supply unit. The conventionally fired cylinders were ignited at the crank angle corresponding to the triggering of the laser Q-switch. The fuelling strategy was controlled by the engine’s own electronic control unit which functioned to keep the air– fuel mixture at stoichiometric (approximately 14.7 times the mass of air to fuel). For each optical plug tested online, the laser energy was initially reduced to find the minimum ignition energy (MIE)


Table 1. Laser beam proprieties for different cavity apertures. Diameter of laser cavity aperture (mm)

Mean average beam radius at 1.4 m from laser (µm)

Standard deviation of beam radius (µm)

Percentage error of beam radius (%)

Laser output divergence (mrad)

Beam quality factor M2

1.3 2 3

1293 1427 1823

35 158 105

2.7 11.1 5.8

0.631 0.798 0.717

2 2.9 4.8

(a) Transmitted energy (after optical plug and plasma) [mJ]

12 10 8 6 4 2 0 0

5 10 15 20 Laser pulse energy into optical plug [mJ]

25

10

60

(b) Transmitted energy (after optical plug and plasma) [mJ]

25 20 15 10 5 0 0

20

30

40

50

Laser pulse energy into optical plug [mJ]

Transmitted energy (after optical plug and plasma) [mJ]

(c) 40 30 20 10 0 0

20

40

60

80

Laser pulse energy into optical plug [mJ] Lens focal length:

15 mm

18 mm

24 mm

30 mm

36 mm

Figure 4. Transmitted energy measured through five optical plugs each with a different FL lens, for three sizes of laser cavity aperture diameter: (a) 1.3 mm, (b) 2 mm and (c) 3 mm.

Figure 5. Variation in the minimum energy required for ignition and air breakdown against lens FL, for the three laser cavity aperture diameters (a) 1.3 mm, (b) 2 mm and (c) 3 mm.

for misfire free combustion. The energy was then increased in increments up to the maximum laser drive level. At each laser energy setting, the steady state values of the COVIMEP and Var PPP for cylinders 1 and 4 were obtained over

300 consecutive engine cycles [15] (24 s duration) with IMEP and PPP values recorded at each cycle. To confirm these values, this process was repeated twice more for each energy setting to obtain average steady state values of COVIMEP and 4733

J D Mullett et al

(a) Laser irradiance (GW/cm2)

200

150

100

50 10

15

20 25 30 Focal length of lens (mm)

35

40

10

15


35

40

10

15


35

40

(b) Laser irradiance (GW/cm2)

300 250 200 150 100 50

(c) Laser irradiance (GW/cm2)

200

150

100

50

Minimum ignition irradiance

Minimum air breakdown irradiance

Figure 6. Variation in the minimum irradiance required for ignition and air breakdown against lens FL, for the three laser cavity aperture diameters (a) 1.3 mm, (b) 2 mm and (c) 3 mm.

Var PPP . Each test was performed at a constant engine speed of 1500 rpm, with each cylinder being fired at 30◦ before top dead centre. The results were only recorded once the engine had warmed up (coolant temperature >80 ◦ C) which was approximately 1000 s after engine start up, to ensure that the conventionally fired cylinders were operating optimally.

3. Results and discussion The cross-sectional beam intensity profiles at 1.4 m from the laser head for the three different cavity apertures can be seen in figure 3. This shows the beams produced by the 1.3 and 2 mm diameter cavity aperture to have a near Gaussian TEM00 mode, 4734

whereas the 3 mm aperture produces the rectangular transverse mode pattern TEM20 . Table 1 shows the mean beam radius for laser energy drive levels between 40% and 100% with the standard deviations and percentage errors, and includes the measured laser output divergence and calculated M 2 for the three cavity apertures. The offline results for increasing the pulse energy into the five optical plugs are illustrated in figure 4 for the different cavity apertures, which shows the transmitted energies through the plugs. The peak of each curve indicates the minimum energy required to cause optical air breakdown at atmospheric pressure for the respective FL lenses. After this point, plasmas were formed which absorbed the incident energy. It can be seen from figure 4 that at the higher plasma producing input energies, the transmitted energy through the optical plugs and plasma becomes fairly constant, which indicates that a percentage of the incident energy is being absorbed by the plasma. However, there will be an energy point where the plasma becomes saturated. This can be seen at the higher energy input end on the graphs shown in figures 4(a) and (b), where the curves start to rise indicating that additional energy is no longer being absorbed by the plasma. This is not shown in figure 4(c) as the energy had to be restricted to a maximum of 80 mJ to avoid damage to the lens and window. It is worth noting that when the optical plug is used online, the proportion of the laser energy that is not absorbed by the plasma would impinge on the piston head and may over time cause undesired damage. This problem may be exacerbated if longer FL lenses are used to focus the beam, as a greater amount of energy is needed to cause breakdown, and only a portion of that energy is used to sustain the plasma, as shown in the offline results in figure 4. Moreover, the beam divergence after the focal point would be less for longer FL lenses and would therefore produce a greater laser irradiance on the piston due to a smaller beam spot. However, the laser irradiance on the piston head would vary depending on a number of variables such as laser energy, FL of lens, focal point location, beam quality, spark timing and cylinder geometry. This is therefore an issue which should be addressed in future research. As seen in previous LI experiments at The University of Liverpool (up to 2 h continuous operation), inspection of the optical plug window after the online engine testing indicated that the deposition of particulates from combustion was only evident in areas of the window that were not irradiated by the laser beam. This suggests that a self-cleaning (thermal ablation) mechanism was active during combustion, which has also been observed and studied in other work in the field [7, 9, 19, 20]. A suggested theory for this mechanism is that, during the first instant of the laser pulse duration, the combustion contamination present on the window partially absorbs the high intensity laser energy and the particulates are rapidly heated and ablated from the surface, thus allowing the remaining pulse energy to propagate to the focal point within the combustion chamber and create the ignition spark. The minimum energies needed to produce dielectric breakdown in air at atmospheric pressure for the different FL lenses used in the optical plugs can be seen in figure 5, along with the MIEs required for misfire free LI combustion obtained from the online results. It is evident that more energy


(a)

(b) 2.0 Normalized: LI divided by SI

Normalized: LI divided by SI

2.5 2.0 1.5 1.0 0.5 0.0

1.0

0.5

0.0 2

4 6 8 10 12 14 Beam energy per pulse in cylinder (mJ)

16

2

(c)


16

(d) 2.5

3.0 Normalized: LI divided by SI


1.5

2.0 1.5 1.0 0.5 0.0

2.5 2.0 1.5 1.0 0.5 0.0

4

6

8

10

12

14

16

6

Beam energy per pulse in cylinder (mJ)

8

10

12

14

16

Beam energy per pulse in cylinder (mJ)

(e) Normalized: LI divided by SI

2.0

1.5

1.0

0.5

0.0 8

10

12

14

16

18

Beam energy per pulse in cylinder (mJ) COVIMEP LI / COVIMEP SI

VarPPP LI / VarPPP SI

Figure 7. Effects of increasing laser energy in cylinder 1 on the ratios of COVIMEP of LI/COVIMEP of SI and Var PPP of LI/Var PPP of SI, for a 1.3 mm laser cavity aperture diameter and lens FLs of: (a) 15 mm, (b) 18 mm, (c) 24 mm, (d) 30 mm and (e) 36 mm.

is required for ignition/breakdown using longer FL lenses, due to the larger minimum waists produced. In addition, figure 5 shows that in general, the energies required for ignition in the engine cylinder are lower than for air breakdown, which is most probably due to the higher pressures and temperatures. The lowest MIEs for the online LI experiments for the 1.3 mm, 2 mm and 3 mm diameter cavity apertures were found to be 3.9 mJ, 10.2 mJ and 12.3 mJ respectively, from using the 15 mm FL lens in the optical plug. These energies are much lower than the energies used by standard ignition coils, which are in the range 30–50 mJ [15]. Even the greatest MIE of 28.3 mJ, found from using 36 mm FL lens with the 3 mm laser cavity aperture, is below this range.

The minimum laser irradiances required for LI and for air breakdown for the different FL lenses were calculated and can be seen in figure 6 for the three laser cavity apertures. Like the minimum energies, the laser irradiance levels required for ignition are lower than for air breakdown. The 1.3 mm aperture results illustrated in figure 6(a) show that the irradiance required for both ignition and air breakdown decreases with longer FL lenses. This is most likely due to the fact that longer FL lenses have greater Rayleigh ranges (depth of focus) and hence have greater focal point volumes. Consequently, the maximum laser irradiance at the focal point will interact with a larger volume of fuel, therefore increasing the probability of breakdown. The minimum irradiance will however reach 4735

J D Mullett et al

(a)

(b) 2.0 Normalized: LI divided by SI


3.0 2.5 2.0 1.5 1.0 0.5

1.0

0.5

0.0

0.0 5


40

(c)

10

15 20 25 30 Beam energy per pulse in cylinder (mJ)

35

15

20 25 30 Beam energy per pulse in cylinder (mJ)

35

(d) 3.0 Normalized: LI divided by SI

2.5 Normalized: LI divided by SI

1.5

2.0 1.5 1.0 0.5 0.0

2.5 2.0 1.5 1.0 0.5 0.0

10


45

(e) Normalized: LI divided by SI

3.0 2.5 2.0 1.5 1.0 0.5 0.0 20

25 30 35 Beam energy per pulse in cylinder (mJ) COVIMEP LI / COVIMEP SI

40

VarPPP LI / VarPPP SI

Figure 8. Effects of increasing laser energy in cylinder 1 on the ratios of COVIMEP of LI/COVIMEP of SI and Var PPP of LI/Var PPP of SI, for a 2 mm laser cavity aperture diameter and lens FLs of: (a) 15 mm, (b) 18 mm, (c) 24 mm, (d) 30 mm and (e) 36 mm.

a threshold limit as lens FL increases, as there will be a threshold value at which dielectric breakdown of the air–fuel mixture occurs. Therefore, at this irradiance threshold, the MIE required would become exponential as lens FL increases. The 2 and 3 mm diameter cavity aperture results illustrated in figures 6(b) and (c) differ slightly from the 1.3 mm cavity results (figure 6(a)), in that the minimum air breakdown irradiance has a greater scatter over the FL range. Although, the averaged trendlines show that the minimum air breakdown irradiances are fairly constant for the range of lens FLs at approximately 250 GW cm−2 for the 2 mm aperture and 140 GW cm−2 for the 3 mm aperture. A reason why these minimum irradiance values do not conform as well as the 4736

1.3 mm aperture results is that the beam size at the optical plug has a greater variance over the laser energy drive range, which would affect the minimum spots produced. This is shown in table 1, where the percentage errors of the standard deviation for the mean beam radius for the 2 mm and 3 mm apertures are 11.1% and 5.8%, respectively, compared with a low 2.7% for the 1.3 mm aperture. To directly compare the performance of the laser ignited cylinder with the conventionally ignited cylinder, the COVIMEP and Var PPP values were normalized by dividing the COVIMEP and Var PPP for LI by the COVIMEP and Var PPP for SI, respectively, to obtain a ratio for each of the optical plug lenses; whereby any ratio values