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

Подождите немного. Документ загружается.

4.1 Application of Lean Flames in Internal Combustion Engines 263

previous generations of so-called wall-guided stratiﬁed-charge design (see following

discussion). The importance of the latter observation is that fuel consumption is

reduced due to stratiﬁed-charge DI not only over a large part of test drive cycles,

but also in ‘real-world’ driving. One of the determinants for mode switching to

homogeneous stoichiometric is governed by when the temperature of the exhaust

gases exceeds the active temperature window of the NO

storage converters, at about

500

◦

C[63]. It is on this basis that it is claimed that SI stratiﬁed DI engines have the

greatest single potential for a reduction in fuel consumption compared with all other

concepts [63]. Once again, however, the reduction in NO

emissions that is due

to even such ultra-lean-burn conditions is insufﬁcient to meet emissions legislation

and elaborate, and costly, NO

after-treatment measures are required for lean-burn

operation. Both engines [63, 64] are reported to have an ‘external’ EGR system to

reduce production of NO

in stratiﬁed mode.

Because of these and other advantages, research and development into DI gaso-

line engines has a long history, dating from at least their use in aeroplane engines in

the 1930s and in automobile racing engines since at least the 1950s. Stratiﬁed systems

also have been researched since at least 1950. However, mass-production engines

using DI and lean-burn combustion appeared only in the mid-1990s, in part because

of the impetus of legislation relating to CO

emission and in part because important

enabling technologies became available. These included

advances in high-pressure common-rail system fuel injector technology (to be

an accumulator and damper, initially at less than 12 MPa, and to be both reliable

and have low parasitic loss), which enabled the generation of ﬁnely atomised fuel

in the combustion chamber and to be able to inject the r equired fuel quantity

into the cylinder; and

development of control technology that enabled the selection of appropriate

injection timing to achieve certain types of combustion and to track the rapid

changes in load [55, 65–67].

It is useful to identify six processes associated with DI engines, whether stratiﬁed or

not, namely fuel injection, spray atomization, droplet evaporation, charge cooling,

mixture preparation, and the control of in-cylinder motion. DI SI engines are now

classiﬁed by the last two processes: There are three methods by which fuel vapour

from the evaporating spray is convected to the spark plug. A schematic of these

methods, namely the wall-guided, air-guided, and spray-guided methods, is shown in

Fig. 4.9 [66].

These methods lead to engine designs that can operate at an overall AFR of up

to about 50 (or equivalence ratio, φ = 0.25) by stratifying the charge while retaining

a reliably ignitable mixture (that is, close to stoichiometric) in the vicinity of the

spark plug. This is achieved, in principle, by retarding the start of injection during

the compression stroke until it is quite close to the ignition point, by designing

an injector with an appropriate spray pattern and with sufﬁciently ﬁnely atomised

droplets to evaporate before ignition, and by control of the ﬂow in the combustion

chamber. The wall-guided concept generates bulk convection to transport the fuel

spray or, better, its vapour, reliably to the vicinity of the spark plug at the time

of ignition. Subsequently the mixture must allow stable propagation of the ﬂame

through the increasingly lean vapour cloud. Even from this brief description, the

264 Lean Flames in Practice

Exact positioning of spray

and spark plug required

High stress on spark plug

Fuel transport to spark plug

due to internal flow

Specification of spark plug

remains standard

Actual development

to wall-/air-

guided systems

(a) (b) (c)

Figure 4.9. Classiﬁcation of DI systems for SI engines: (a) spray-guided, (b) wall-guided,

need is clear for precise management of the intake port geometry, of the in-cylinder

ﬂow, especially the proﬁle of the piston crown, and its interaction with the spray, and

hence of the spray’s position and direction. These aspects of design are essential for

a successful operation over the requisite wide range of speeds and loads. Broadly,

the required ‘endpoint’ is to have [59] a slightly rich mixture at the spark-plug

electrodes at ignition, a near-stoichiometric mixture surrounding the initial ﬂame

kernel to facilitate ﬂame propagation, and steep gradients at the edge of the fuel

cloud to minimise the amount of fuel that is below the ﬂammability limit and that

will give rise to unburnt HC emissions. As Drake and Haworth [59] point out,

the management process must also take into account the effects of cycle-by-cycle

variations to produce a robust design.

Fuel Premixing Upstream of the Inlet Valves

Prechamber design. Figure 4.10 shows an early four-stroke passenger-car SI engine

that made use of lean-burn combustion for the purpose of lowering emissions: The

design [68] was among the ﬁrst ‘stratiﬁed-charge engines’ in regular production [53],

at least outside the former Soviet Union (there, designs of engines based on so-

called LAG combustion (meaning avalanche activated combustion) [69], an early

Auxiliary intake

valve

Auxiliary intake

passage

Spark plug

Main intake

valve

Figure 4.10. Prechamber stratiﬁed-charge engine

with prechamber inlet valve and auxiliary carburet-

tor (E. Taguchi, personal communication).

4.1 Application of Lean Flames in Internal Combustion Engines 265

Figure 4.11. Four-cylinder, four-valve, PFI stratiﬁed-charge automobile engine with variable

valve control system (VTEC-E) [70].

form of HCCI engine, as subsequently described below, were developed based on

Nobel Laureate Semenov’s work on chain branching in chemical reactions). The

purpose of the design was to make use of the low-emission attributes of lean burn

combustion to meet the legislative requirements responding to increased awareness

of the noxious effects of pollutant emissions such as CO, NO

, and unburnt HCs,

most notably in the Clean Air Act Extension of 1970. However, the speciﬁc fuel

consumption of the engine was also good for its time, while avoiding the need for

any catalytic conversion. The design was an example of a ‘divided chamber’ or

‘prechamber’ that provided jet ignition to the lean mixture in the main combustion

chamber. A more recent prechamber design was described in [5].

PFI design. As legislative requirements became more stringent, a four-stroke, four-

valve, PFI, stratiﬁed-charge engine eventually superseded this prechamber design

[70], shown in Fig. 4.11 and with the associated speciﬁcations in Table 4.2. Figure 4.12

shows the qualitative features of the ﬂow ﬁeld when the VTEC-E (variable valve

timing and lift electronic control – economy) mechanism activates only one of the

two intake valves with appreciable lift; the mechanism itself is shown schematically

in Fig. 4.13. The variable valve timing and lift mechanism consists of one camshaft

with two cam proﬁles: one, denoted as the primary cam, is permanently engaged and

produces a full lift in the corresponding valve. The other, secondary, cam produces

a maximum lift of only 0.65 mm, as shown in Fig. 4.14, thereby deactivating it. The

266 Lean Flames in Practice

Table 4.2. Speciﬁcation of variable valve timing and

lift electronic control–economy (VTEC-E) engine

Cylinder conﬁguration In-line cylinder

Bore × stroke (mm) 75 × 84.5

Displacement (cc) 1493

Compression ratio 9.3:1

Valve train SOHC

†

VTEC-E

Number of valves 4

Inlet-valve diameter 27.5

Exhaust-valve diameter 23.5

Fuel injection system PGM

∗

-FI

Ignition system PGM-IG

†

SOHC: single overhead camshaft.

∗

PGM: programmed fuel injection.

primary rocker arm has a roller follower, to reduce friction, with a built-in hydraulic

piston. At low engine speeds, each of the two rocker arms works independently of

the other according to the corresponding cam proﬁle. At high speeds, the hydraulic

piston engages the rocker arm of the secondary valve so that both valves are fully

lifted. This engine has three speciﬁc features:

1. At high engine speeds, it prevents the decrease of the volumetric efﬁciency by us-

ing straight ports and activating both intake valves. At these speeds, the charge

Figure 4.12. VTEC-E variable valve control sys-

tem showing primary and secondary rocker arms in

the foreground: The swirl component is generated

at low speeds and loads [70].

4.1 Application of Lean Flames in Internal Combustion Engines 267

High Speed

(Two-Intake-Valve Operation)

Low Speed

(One-Intake-Valve Operation)

Oil Pressure

1 Primary Rocker Arm

2 Secondary Rocker Arm

3 Hydraulic Piston

4 Oil Passage

5 Roller Follower

6 Primary Cam

7 Secondary Cam

8 Rocker Shaft

Figure 4.13. Mechanical and hydraulic

actuation of the variable valve at low and

high speeds [70].

is stoichiometric and changes to lean burn at lower speeds. With few excep-

tions, such mode switching has been the hallmark of most lean burn combustion

systems.

2. At low engine speeds, it generates mean swirl, as well as mean tumble, in the

cylinder by means of a mechanism that deactivates one of the two intake valves,

producing a complex overall mean ﬂow pattern [71]. The swirl and tumble ratios

were about 2.5 and 1.2, respectively;

3. At low engine speeds, it produces a lean-burn axially stratiﬁed charge by con-

trolling the fuel injection timing in the intake stroke.

The second feature, and the generation of mean tumble in particular, is impor-

tant in the following context. Lean mixtures have a slower burning rate than do

Exhaust

Crank Angle (degree)

Valve Lift (mm)

−180

(BDC)

(TDC)

180

(BDC)

Intake

Primary Valve

Secondary Valve

Figure 4.14. Lift proﬁles of the primary and secondary valves, the latter during low-speed

operation [70]. BDC, before dead centre; TDC, top dead centre.

268 Lean Flames in Practice

57 6

1 Test engine with transparent piston

2 CCD camera

3 CCD camera controller

4 Image processor

5Ar

Laser

6 Light separator

7 Acoustic optical modulator

8 Optical fibre

9 Laser-sheet optics

10 Crank angle signal amplifier

11 Trigger controller

12 Acoustic optical modulator driver

13 Tracer feeder

Figure 4.15. Schematic of the particle coded-pulse velocimeter [73]. CCD, charge-coupled

device.

stoichiometric ones, and this can lower the fuel-conversion efﬁciency of the cycle as

well as increase exhaust emissions. The burning rate can be increased by promoting

the scale and magnitude of the turbulent velocity ﬂuctuations during combustion, and

[5] describes the use of squish-jet motion, produced by the geometry of a specially

design piston crown, to achieve these ends. Here it is likely that the conventional

description [72] of generating turbulence near and after TDC is the important mech-

anism. Arguments based on the conservation of angular momentum suggest that the

velocities associated with the tumble vortex motion should increase by a factor re-

lated to the compression ratio of the engine. A particle coded-pulse velocimeter [73],

shown in Fig. 4.15, made 3D measurements of velocity in an optically accessed en-

gine. The measurements are shown in Fig. 4.16. The measurement plane was 16 mm

below the spark plug, illuminating a section across the bore, and the engine speed

was 1500 rpm: 450

◦

CA (crank angle) corresponds to the intake ﬂow whereas the

other three ﬁgures show the ﬂow during the compression stroke. The colour con-

tour (see colour plate) shows the tumble (axial) component of velocity: red shows

axial velocity components ﬂowing from piston crown to the spark plug: Blue shows

the opposite direction. The arrows show the swirl ﬂow component. It is clear that

the tumble component velocity magnitudes remain unchanged during compression.

Thus the additional energy that is due to compression that is, in principle, put into

the tumble vortex (as the density of the gases increase) is converted, by means of

tumble vortex breakdown, into turbulent kinetic energy. Also, in agreement with ex-

pectation [72], the magnitude of the swirl component of velocity remains unaltered

during compression.

The effect of injection timing on the NO

emissions, on the co-variance in the

indicated mean effective pressure (IMEP), and on the normalised AFR ratio at the

spark plug (measured by a fast-acting sampling valve) as a function of the start of

injection are shown in Fig. 4.17. The overall AFR was maintained at 23. The vertical

4.1 Application of Lean Flames in Internal Combustion Engines 269

Cylinder Bore

(a) 450° CA

(b) 540° CA

660° CA

exhaust

intake

Primary Valve

Secondary Valve

Vh: 20m/sec

Measurement Plane:

16mm below the spark plug gap

Timing:

ignition timing (BTDC40*)

V axial (m/sec)

Figure 4.16. Three components of phase-averaged velocity at four crank-angle (CA) posi-

tions measured by the particle coded-pulse velocimeter. Note that the diameter of the ﬁeld of

view through the piston crown is less than the bore of the engine. Firing TDC is at 720

◦

[73]. (See colour plate.)

band shows the optimum fuel injection timing during the intake stroke that enables

quite stable combustion with a relatively low NO

emission level. The questions

thus arise as to how this rapid variation in the AFR arises with injection timing

and whether further development could result in the engine running at even higher

overall AFRs.

Although these measurements show how some aspects of the lean burn come

about, the mechanism whereby injection upstream of open inlet valves leads to ade-

quately stratiﬁed charge at the spark plug, after almost 360

◦

of crankshaft revolution,

is not clear. Extensive measurements by phase-Doppler anemometry of droplet size,

velocity, and ﬂux were carried out. These showed that gasoline injection against an

open inlet valve controlled stratiﬁcation through the amount of liquid phase enter-

ing the engine. The droplets evaporated at or near the cylinder liner, and gas-phase

velocity measurements suggested that the resulting fuel-rich cloud was convected

by the tumble vortex initially down towards the piston crown and subsequently

back towards the cylinder head and into the pent-roof. Associated measurements of

the fuel vapour within the cylinder measured by planar laser-induced ﬂuorescence

(PLIF) [74] conﬁrmed this view of events. Figure 4.18 shows measurements at 1500

rpm, by PLIF, of the mixture distribution in the pent-roof, 6 mm below the spark

plug at ignition [40

◦

CA before top dead centre (BTDC)]. The light source was

a XeCl (308-nm) laser sheet of 1-mm thickness, iso-octane was used as fuel and

mixed with 10%vol/vol of 3-pentanone as the s eed. A cooled charge-coupled device

(CCD) camera with an image intensiﬁer observed the ﬂuorescence intensity: The

mixture distribution strength is coded by colour, red showing t he richer areas and

270 Lean Flames in Practice

270 360(TDC)

Δ (A/F ) (%)

ISNO

(g/kWh)COV

IMEP

(%)

in[(s)

(° )

450 540

1500 rpm

= 34%

A/F = 23

= MBT

−5

rich

lean

−10

Figure 4.17. Inﬂuence of the timing of the start

of injection on indicated speciﬁc NO

emissions,

the covariance (COV) of the IMEP, and on the

variation of normalised AFR. Intake TDC is at

360

◦

CA [70]. MBT is minimum advance for best

torque.

green showing the leaner ones (see colour plate). As the timing was changed from

injection against a closed valve [Fig. 4.18(a), 0

◦

CA] to against an opening valve

during the induction stroke [Fig. 4.18(b), 20

◦

CA after top dead centre (ATDC) and

Fig. 4.18(c) 55

◦

CA ATDC], a richer mixture appeared near the spark plug. Detailed

Exhaust

Spark plug

Intake

(a) Injection = 0° CA

Measurement Plane:

8mm below the spark plug gap.

Timing:

ignition timing (BTDC40°)

Valve lift and timing for VTEC-E mode

(b) Injection = 20° CA (b) Injection = 55° CA

Start of Intake Stroke

Middle of Intake Stroke

Exhaust Intake

Primary valve

Valve lift (mm)

Secondary valve

Chank Angle (deg.)

-180

(BDC)

(TDC)

180

(BDC)

Figure 4.18. Contours of tracer PLIF measurements of fuel concentration near the spark-

plug electrodes as functions of the start of injection timing (

◦

CA afer TDC). Red is relatively

richer, yellow is relatively leaner. Intake TDC is at 0

◦

CA. (See colour plate.)

4.1 Application of Lean Flames in Internal Combustion Engines 271

modiﬁcations to the design of the intake ports and to the fuel injection equipment

were able to improve the limits of the design, for reasons described by Carabateas

et al. [75], who performed measurements similar to those of [73]. To some extent, it

remains surprising that a reasonably compact, relatively fuel-rich cloud can survive

from just after TDC of inlet to TDC of compression. It may be that the suppression

of turbulence by the swirl proﬁle is an important near-wall mechanism [72].

As the AFR of the mixture is increased above stoichiometric, cycle-by-cycle vari-

ations in ﬂame development and consequently in the IMEP become more prominent

and eventually limit the range of lean-burn operation because these cause poor drive-

ability. Therefore it is interesting to understand how, at least in principle, to extend

the limits of the AFR at which engine designs operate. A critical stage of combus-

tion in SI engines is during initial ﬂame-kernel growth that corresponds, typically, to

the duration of 0%–5% mass fraction burned (MFB). In particular, the correlation

between the IMEP and the CA corresponding to 5% MFB (θ

Xb5%

) has been found

to be quite strong under low-load lean-burn conditions [76], although the ﬁgure of

5% is arbitrary in the sense that it was determined by experimental convenience:

It is hard to measure lower values of MFB from the indicator diagram because of

uncertainties. Imaging of the initial ﬂame-kernel growth has proven particularly use-

ful in investigating combustion behaviour at CA degrees closer to ignition timing

than θ

Xb5%

, corresponding to less than 0.1% MFB. The studies in [76–78] were con-

cerned with investigation of such early stage ﬂame-kernel activity by use of either

stereoscopic or double CCD imaging per cycle. These images suggested that large

variations affect the initial ﬂame-kernel growth, either through stretch, wrinkling,

inhibition of growth by wall proximity, or speed of convection towards, or away

from, the electrodes. Surprisingly, and disappointingly, their ﬁndings suggested that

the effects of ﬂame convection, ﬂame distortion, and heat losses to the electrodes

were not the major contributors to cyclic variability. The differences between these

quantities for very fast and very slow cycles were quite small. However, it was found

that ﬂames from fast cycles had occasionally shape properties similar to the ones

of stoichiometric ﬂames and a ﬂame-kernel growth modelling approach also sug-

gested that the level of cyclic variability in burning rates could be explained by

large variations in the AFR. In subsequent work, [79] reported an experiment using

chemiluminescence to evaluate the in-cylinder equivalence ratio of the reacting mix-

ture and to further examine its contribution to the ﬂame-growth speed and the cyclic

variability in the CA by which 5% mass fraction was burned (θ

Xb5%

). The results

showed that the equivalence ratio exhibited large variations on a cycle-by-cycle ba-

sis and consistently produced negative correlation coefﬁcients with θ

Xb5%

, especially

for lean set-operating conditions (AFR = 20–22). For injection against open inlet

valves that yielded a stratiﬁed mixture, the degree of this correlation was in the range

from about −0.4 to −0.8. Whether the level of cyclic variability in the fuel-droplet

distribution patterns during the early intake stroke could account for the variations

in the equivalence ratio was examined by Aleiferis et al. [80]. Their ﬁndings, based

on light scattering from a planar laser sheet, showed consistent differences in the

intensity scattered between poorer-than-average and better-than-average values of

the IMEP. However, further work is needed to establish whether the larger scattered

intensity associated with better-than-average cycles was due to a larger number of

droplets or to larger droplets.

272 Lean Flames in Practice

Deep-bowl-type piston cavity

Spark plug

High-pressure swirl injector

Figure 4.19. The i-VTEC I combustion

chamber conﬁguration [82].

Nevertheless, it is clear that PFI designs give rise to too much time between

injection and ignition, owing to the distance between the injector and the spark plug,

and too much turbulent mixing to hope that combustion at much leaner overall

AFRs can be achieved. Fortunately, the ability to bring the injector and the spark

plug closer together, and thereby burn at substantially leaner overall AFRs, began

to be realised at commercial levels of robustness in the mid-1990s through DI of fuel

into the combustion chamber, as described under the next heading.

Fuel Premixing Using DI

Wall-guided stratiﬁed DI concepts (‘ﬁrst generation’). Wall-guided designs were ex-

tensively developed and analysed by optical techniques, as shown by the comprehen-

sive investigation [81] used to develop the ﬁrst ‘modern’ commercial LBDI engine.

We use the engine described in [82], one of the most recent designs to have used the

stratiﬁed wall-guided approach, as an illustration of some of the design objectives

that a stratiﬁed LBDI engine must achieve. The engine was based on the control

of the burn rate by the VTEC mechanism (the same as the VTEC-E design previ-

ously described for the PFI) through the in-cylinder swirling motion of the gas. The

engine is shown in Fig. 4.19, with the associated speciﬁcations in Table 4.3. In this de-

sign, a high-pressure atomiser was located in the centre of the combustion chamber

with the fuel spray injected vertically downwards into the cylinder, towards a piston

bowl, with the spark plug placed between the exhaust valves. As brieﬂy described in

the following discussion, optimisation was required of the mixture preparation and

combustion through calibrations of the piston bowl, of the fuel-spray characteristics,

and of the in-cylinder gas motion to enable stable combustion with an ultra-lean

AFR. This centre-injection location eased the constraints on the timing of injection,