Swaminathan N., Bray K.N.C. (eds.) Turbulent Premixed Flames

Подождите немного. Документ загружается.

4.1 Application of Lean Flames in Internal Combustion Engines 273

Table 4.3. Speciﬁcation of i-VTEC I engine

Engine model code K20B

Cylinder conﬁguration In-line 4-cylinder

Bore × stroke (mm) 86 × 86

Displacement (cc) 1998

Compression ratio 10:1

Valve train DOHC

VTEC

Number of valves 4 per cylinder

Inlet-valve diameter 34

Exhaust-valve diameter 28

Fuel injection system Programmed fuel injection

Fuel supply pressure (MPa) 8–10.5

Fuel Regular (91 RON

∗

)

Max. power (kW/rpm) 115/6300

Max. torque (Nm/rpm) 188/4600

Double overhead camshaft.

∗

Research octane number.

imposed by the injector and piston cavity geometry, found with the other choice for

injector location, namely side injection. This made it possible to optimize injection

timing for any operating condition and to achieve stratiﬁed combustion across a wide

range of the operation map. However, in common with most engines of this design,

stratiﬁed combustion was used on only a portion of one of the two VTEC-deﬁned

modes of the operation map; that is, only for the low loads and at low speeds. The

general relationship between the brake-speciﬁc fuel consumption (BSFC) and the

engine displacement is shown in Fig. 4.20. As the result of the stratiﬁed lean com-

bustion, this engine achieved a 20% improvement in BSFC compared with that of

engines of the same displacement under stoichiometric fuelling through the effects

of reduced heat losses, increased speciﬁc heat ratio, and reduced pumping losses.

To meet the Japanese emissions regulations for 2005, during ultra-lean com-

bustion, EGR was introduced. Figure 4.21 shows the inﬂuence of the EGR rate, at

an engine speed of 1500 rpm and 118-kPa brake mean effective pressure (BMEP),

on BSFC, engine-out NO

emissions, and combustion stability: Note that the AFR

could be as lean as 65:1 on this engine. The engine operating point that provided the

550

500

450

400

350

300

1500 1700 1900

Displacement (cm

)

2100 2300

20% improvement

Stoichiometric

engines

Port-fuelled lean-burn

engines

LBDI engine

2500

BSFC (g/kWh)

Figure 4.20. Comparison among BSFCs

of PFI stoichiometric, PFI lean-burn and

LBDI engines at 1500 rpm and a BMEP

of 118 kPa. Modiﬁed from [82].

274 Lean Flames in Practice

01020

EGR Rate (%)

BSFC (g/kWh)

Co-variation

of IMEP (%)

BSME

(g/kwh)

AFR(–)

30 40 50

420

410

400

390

380

Figure 4.21. Inﬂuence of EGR on BSFC

and engine-out emissions at 1500 rpm and

a BMEP of 118 kPa. Modiﬁed from [82].

BSME is brake speciﬁc mean emission.

minimum BSFC had an EGR rate as high as 40% and yet it also has achieved sta-

ble combustion. In addition, this high EGR rate reduced engine-out NO

emissions

by 80% compared with the case without EGR, although the introduction of EGR

slightly reduces thermodynamic efﬁciency because of its polytropic index and has,

on occasion, been reported as increasing the emissions of hydrocarbons because of

quenching. Together with an efﬁcient lean NO

catalytic converter, a vehicle with

this LBDI engine achieved emissions at 25% of the level stipulated by the Japanese

emissions regulations and was the ﬁrst to obtain an ‘ultra-low-emission vehicle’

(ULEV) certiﬁcation. To achieve this performance, higher conversion efﬁciencies

of HCs during warm-up with stoichiometric operation, as well as during lean-burn

operation, were realized through the precise control of AFR by use of a heated

exhaust-gas oxygen (HEGO) sensor downstream of a three-way catalytic converter

and a universal exhaust-gas oxygen (UEGO) sensor upstream from the converter.

A high-efﬁciency lean NO

catalyst with a conversion efﬁciency of about 98%–99%

at around 300–350

◦

C of catalyst temperature was also installed. The vehicle was

capable of a fuel economy of 15.0 km/L in the 10–15 Japanese driving cycles, which

is 15% higher than the fuel-economy standard for the year 2010.

At high-power output, switching to two-intake-valve operation – with the intake

stroke injection providing a homogenous stoichiometric mixture – increased the

volumetric efﬁciency. Furthermore, the choice of injection geometry avoided wall

wetting of the fuel on the cylinder wall, enabling reduction of unburned HCs (UHCs).

4.1 Application of Lean Flames in Internal Combustion Engines 275

1 10 100 1000

Droplet size (μm)

Frequency (%)

Direct injector

(10 MPa)

SMD-11, 4 μm

Port injector

(400 kPa)

SMD-11 1 μm

Figure 4.22. Comparison of SMDs be-

tween conventional port injector and

high-pressure direct injector [82].

At low loads and speeds, to achieve ultra-lean stratiﬁed combustion, generating the

adequately rich mixture clouds near the spark plug and achieving fast combustion

were important. For mixture formation, it was better to restrain the in-cylinder

ﬂow to the extent possible. On the other hand, to realize rapid combustion, which

improves thermal efﬁciency, a high turbulence intensity was required. To satisfy these

opposing requirements and for the formation of stratiﬁcation in the centre-injection

combustion concept, the injector had to have a narrow spray angle, here set to 25

◦

and a spray that was capable of high penetration. The injection pressure was set at

10 MPa and the resulting Sauter mean diameter (SMD) of the fuel droplets was

smaller than about 12 μm; see Fig. 4.22. Rapid combustion was obtained through

the swirl generated by deactivation of one of the intake valves, using the VTEC

mechanism.

To develop an efﬁcient design of a combustion chamber, a commercial com-

putational ﬂuid dynamics (CFD) code (AVL “Fire” Version 7.1) was used, with

experimental veriﬁcation of its reliability of prediction of the mixture formation in

the cylinder by comparison with experiment. Optical access through a glass piston

crown provided measurements of the mixture preparation by use of PLIF, as de-

scribed earlier. The instrument layout is shown in Fig. 4.23. The laser light sheet was

introduced through the piston crown, and the mixture preparation in a vertical plane

was observed under ﬁring conditions. The comparison between PLIF experimental

results and CFD estimations is shown in Fig. 4.24 for the particular conditions of an

engine speed of 1500 rpm, an IMEP of 267 kPa, and an injection timing of 40

◦

BTDC.

A mesh was generated, with small amendments by hand, of between 50 000 cells

at TDC and 90 000 cells at bottom dead centre (BDC) with an integration time

step of 0.2

◦

CA. Equations were solved in Reynolds-averaged form by use of the

k–ε turbulence model, with boundary conditions at the valves provided by a duct

acoustic simulation code (AVL BOOST). The wall temperature was ﬁxed during

the whole cycle. A richer-than-average mixture was found around the s park gap in

both results, as well as on the opposite side of the cylinder bore from the spark plug,

and the development of the mixture formation processes as functions of the CA

was well matched. Figure 4.25 shows CFD results of the in-cylinder ﬂow and the

mixture preparation during stratiﬁed combustion. The penetration of the injected

spray of the high-pressure fuel generated a toroidal vortex within the piston cavity,

276 Lean Flames in Practice

II-CCD

Injector

driver

Trigger

controller

CCD camera

controller

Excimer laser

(Xacl 308mm)

Laser

sheet optics

Crank angle

signal AMP

Injection

controller

Figure 4.23. PLIF measurement system for visualisation of mixture formation [82].

Figure 4.24. In-cylinder mixture formation: Comparison between PLIF and CFD as functions

of the CA of the end of injection. Modiﬁed from [82]: Scale for equivalence ratio applies to

CFD quantitatively, to experiment qualitatively. PLIF, planar laser-induced ﬂuorescence. (See

colour plate.)

4.1 Application of Lean Flames in Internal Combustion Engines 277

Mixture preparation Flow field

BTDC 40deg.

BTDC 25deg.

BTDC 15deg.

Figure 4.25. CFD predictions of in-cylinder mixture preparation [82]. (See colour plate.)

resulting in the required compact mixture cloud. The fuel spray injected was rapidly

vapourised by impact with the piston wall, and the mixture was directed to the spark

plug with the ﬂow generated by the spray itself. At an ignition timing of 15

◦

BTDC

there was an ignitable mixture near the spark plug, and as a result even ultra-lean

stratiﬁed combustion at an overall AFR of 65:1 could be reliably ignited.

Figure 4.26 shows combustion stability, engine-out HC, NO

emissions and in-

dicated speciﬁc fuel consumption (ISFC) as a function of the injection timing during

stratiﬁed-charge operation at an engine speed of 1500 rpm and a BMEP of 118 kPa.

The overall AFR was about 53 but nevertheless stable combustion across a wide

range was achieved with an injection timing between 20

◦

BTDC and 50

◦

BTDC.

The optimal injection timing for fuel economy, emissions, and combustion stability

was found to be 40

◦

BTDC. When the injection timing was advanced close to 60

◦

BTDC, the mixture diffused, weakening the stratiﬁcation and the co-variance of the

indicated mean effective pressure (COV IMEP) became worse. On the other hand,

if the injection timing was retarded to near 20

◦

BTDC, the ISFC became worse. As

the HC emissions increased in this scenario, there was probably an increase in late

combustion and the delay in the combustion period caused the deterioration in the

ISFC. The settings for injection timing were thus optimised for a good balance of

emissions and ISFC across engine speeds and loads.

In other wall-guided stratiﬁed-charge DI engines, [59] reports on the emissions of

soot from ‘pool ﬁres’ on the piston surface, caused by the impact of the liquid fuel and

its high-boiling-point components. Further consideration of such ‘rich-burn’ ﬂames

is outside the scope of this contribution, other than to note that smoke emissions

are another restriction on the speed/load range for stratiﬁed engine operation. The

ignition and combustion, as well as soot formation, of the premixed charge were

examined through spectroscopic examination of the emission from the CN molecule,

through OH* chemiluminescence, and using two-colour pyrometry respectively [83,

84]. The ﬁndings of [83, 84] are that optimum injection and ignition timing lead to

an ensemble-averaged equivalence ratio at about φ = 1.5, the bias to a rich average

278 Lean Flames in Practice

Figure 4.26. Inﬂuence of end-of-injec-

tion timing on ISFC and engine-out emis-

sions at 1500 rpm and a BMEP of 118 kPa

[82].

helping to obviate misﬁres (which occur only for φ<0.5) and also giving rise to

tolerance of the large amounts of EGR. In addition, long-duration high-energy

sparks are advantageous for reliable ignition of stratiﬁed mixtures. Combustion

starts as a partially premixed ﬂame that propagates through the ﬂammable mixture.

Sooting is observed in the hot post-ﬂame gases as locally rich regions mix to burnable

proportions, but this soot burns out rapidly because of its high temperature and rapid

mixing with the surrounding hot lean regions.

Spray-Guided Stratiﬁed DI Concepts (‘Second Generation’). The disadvantages of

the wall- and air-guided designs are threefold from the point of view of realising

lean-burn combustion [ 85]; the operating range with lean-burn combustion can be

quite-limited, the efﬁciency potential is reduced by wall wetting and wall heat losses,

and the CA at which 50% of the fuel mass has been burnt may sometimes be

unfavourably phased.

The spray-guided system may be able to avoid these drawbacks. It involves

having a spray discharging more or less along the axis of the cylinder with the

electrodes of the spark plug placed close to the spray edge so as to be able to ignite

the fuel vapour directly, without the need for further convective management by

the in-cylinder ﬂow. This design avoids – in principle at least – contact of the spray

with any wall and thus avoids one source of UHC emission. The fuel cloud can

consequently be smaller, resulting in faster combustion, with fewer cyclic variations

in location or CA and with lower smoke emission. Spray guidance thus delivers [62]

4.1 Application of Lean Flames in Internal Combustion Engines 279

greater potential of the DI stratiﬁed concept, resulting in a wider operation map for

stratiﬁed-charge combustion with improved fuel economy – provided that means can

indeed be found to achieve these objectives successfully over a wide speed and load

range of the engine’s operation map. It is reported that, based on the new European

drive cycle (NEDC) European drive cycle and compared with a wall-guided design,

a spray-guided design had 4% better fuel economy, a factor of two lower UHC

emissions and, as expected, lower smoke emissions [59].

From the preceding qualitative description, it is clear that spark location and

timing relative to that of the spray – and the associated fuel vapour – is critical

[86, 87]. The time available for generation of a vapour cloud conducive to lean burn

combustion is short because the time available for injection and mixing is a few

milliseconds. This implies the need for rapid fuel evaporation and hence droplets of

the order of 20 μm in diameter, which in turn implies the need for development of

a high-pressure (200-bars) gasoline fuel pump, and a design of spark plug open to

the ﬂow of mixture with a suitable – and spatially extended – spark discharge. Thus

the relative orientation of the spark plug, the injector, and of the intake valves is an

important parameter. Evaporation and subsequent mixing should also be rapid in

order to avoid the generation of particulates, and hence this favours the development

of fuel injection equipment operating at yet higher pressures t han required for

producing a ﬁne spray. The UHC emissions can be reduced by reducing the time

between the end of injection and the start of ignition because this results in less time

for the fuel cloud to mix with the surrounding gases to form a too-lean mixture. It is

beneﬁcial to minimise – or preferably eliminate – the wetting of the spark plug by

the liquid fuel (to avoid fouling), while nevertheless reliably establishing near the

electrodes an ignitable vapour mixture at a thermodynamically appropriate ignition

angle over a wide range of engine speeds – and hence wide range of ﬂow speeds and

also ﬂuctuations of fuel vapour – while maintaining stratiﬁcation of the remainder of

the charge to derive the beneﬁts associated with lean burn. It is therefore reasonable

to inquire as to how, where, and when the vapour phase ‘escapes’ from the two-

phase jet for this condition to be satisﬁed, given that the jet will be entraining the

surrounding gases vigorously.

The fuel can be introduced into spray-guided engines by either high-pressure-

swirl injectors, or solenoid-controlled multihole injectors, or piezoelectric ‘outward-

opening’ injectors. Figure 4.27 compares these three designs at atmospheric and

higher back-pressures. For repeatable formation of a lean cloud the cone angle

should, as previously implied, ideally be independent of the back-pressure, and

these particular multihole and piezoelectric injectors fulﬁl this requirement. It is

also desirable that the fuel spray not penetrate too far, to avoid wall wetting, and

the piezoinjector fulﬁls this requirement well. For the multihole injector, the rate of

momentum transfer from the liquid to the surrounding gas is poorer: On the other

hand, the multihole injector is likely to be less affected by the in-cylinder airﬂow

relative to a pressure-swirl atomiser [88]. In addition, some multihole injectors induce

a ﬂow in the surrounding gas that can alter the so-called ‘umbrella angle’ of the

injector [89]: However, other multihole designs can provide better spray targeting

to avoid wall wetting. The rapid mechanical response of the piezoelectric injector

permits multiple injections within the combustion cycle, and it is claimed [85] that

this contributes to, and can extend, stratiﬁed-charge operation by preventing overly

280 Lean Flames in Practice

Figure 4.27. Spray images of a swirl, a multihole, and an outward-opening injector for atmo-

spheric conditions and 0.6-MPa absolute back-pressure [85].

lean or rich areas, the former contributing to ﬂame extinction and HC and CO

emissions, the latter to soot formation. Furthermore, [67] reports that the greater

the quantity of fuel to be injected, in other words the higher the load, the greater the

effect of multiple injection on mixture formation. This observation thereby raises

the possibility of extending lean-burn stratiﬁed operation to higher loads.

As a consequence of measurements by use of optical instrumentation, the funda-

mental processes forming the lean mixture are known, at least qualitatively. There are

extensive parametric measurements of the velocity ﬁeld accompanying the impul-

sively started spray issuing from, respectively, a hollow cone pressure-swirl injector

and from an outward-opening piezoelectric injector [90–92]. For the pressure-swirl

injector type, Fig. 4.28 shows the essence of the mechanism deduced from the mea-

surements from a two-phase PIV instrument at two ambient pressures, the hollow

cone spray collapsing at higher pressures. An inner-vortex structure and an outer-

vortex structure are established by momentum exchange with the spray as it atomises,

and this induced airﬂow is responsible for the structure of the spray as a function

of the ambient pressure. At higher pressure, the entrainment at the outer surface

of the cone is reduced because of the higher droplet density that entails an area

of low pressure inside the spray, followed by the creation of a strong inner vor-

tex. The latter drags the droplets to the nozzle axis, leading to a reduction in the

spray cone angle. The temporal variations of spatial equivalence ratio were mea-

sured [93] and the effect of ambient air entrainment was compared for s ingle and

split injections, as shown in Fig. 4.29. The conclusion is that controlled split injec-

tion can contribute to the optimisation of combustion in DI spark ignition engines.

The phenomenon of high concentration of liquid fuel that is due to droplets ‘piling

up’ at the leading edge of such a spray is circumvented to some extent by use of

appropriate split-pulse injections. Also, the existence of excessively lean mixture in

Injector

Outer vortex

Inner vortex

Air motion

Spray

Injector

Increasing

ambient

pressure

Figure 4.28. Pressure-swirl atomiser: mechanisms of spray–air interaction at different ambi-

ent pressures [90].

Figure 4.29. Pressure-swirl atomiser: Effect of ambient air entrainment on the mixture for-

mations of (a) single injection and (b) split injections at 2.5 ms after start of injection. Injection

pressure, 4.6 MPa; ambient pressure, 0.6 MPa; T = 300 K [93].

282 Lean Flames in Practice

x (mm)

y (mm)

5 101520253035

recirculation vortex

Figure 4.30. Piezoelectric injector: spray propagation and induced gas ﬂow; back-pressure is

6.2 bars, injection pressure is 200 bars, 500 μs after start of injection. The maximum velocities

in the picture correspond to 12 m/s [92]. The vector plots are blended with monochrome

images of the spray, which is visualized with the Mie-scattering technique.

the fuel vapour cloud can be reduced by an appropriate split-injection scheme. The

mechanism by which split injection changes the mixture and spray characteristics is

through the spray-induced motion of the ambient air.

For the outward-opening piezoelectric injector, a broadly similar picture emerges

[92], as shown in Figs. 4.30 and 4.31. As will be subsequently shown, the vortex

structures can be expected to capture some of the evaporated fuel vapour and ‘trap’

it as a more-or-less uniform mixture: So the location of the vortex centres as a

function of back-pressure, of injected fuel mass and, most important, of multiple

injections – which improve operating conditions in stratiﬁed mode – is important.

These aspects were investigated in detail, and it was found that the position of

the two counter-rotating vortices relative to the nozzle is very stable. This leads to

robust engine operation with predictable conditions for mixture formation over

a wide range of operating points. In particular, multiple injections can control the

position and strength of the vortex. This is shown clearly by Skogsberg et al. [94],

who also investigated the development of the velocity and droplet-size characteristics

associated with an outward-opening piezoelectric gasoline injector. Figure 4.32 shows

planar Mie-scattering images at various times after the start of injection with the

counter-rotating torroidal vortical motion shown by superimposed arrows. The upper

parts of the fuel cloud remain in the vicinity of the spark plug well past the time