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

Подождите немного. Документ загружается.

4.1 Application of Lean Flames in Internal Combustion Engines 283

x (mm)

y (mm)

funnel flow

6 8 10 12 14

Figure 4.31. Piezoelectric injector: spray propagation and induced gas ﬂow close to the in-

jector; back pressure is 6.2 bars, injection pressure is 200 bars, 400 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.

corresponding to the end of injection. This should be compared and contrasted

with Fig. 4.33, which shows the corresponding image when the same mass of fuel is

injected in two separate injections. A third vortex, introduced by the second injection,

is formed at the outside edge of the spray. This results in the upper parts of the spray

being pushed back towards the spark plug as a result of the interactions between the

spray-induced vortices: A split-injection strategy can reduce the axial penetration,

a phenomenon that may be useful in moderation to keep a suitable equivalence

ratio near the spark plug – or a nuisance if the droplets strike the injector. Both of

the currently commercially available stratiﬁed-charge engines make use of outward-

opening piezoelectric injectors and both make use of up to three injections per cycle.

A PLIF image of the fuel spray and vapour distribution for a stratiﬁed operating

condition was reported in an optical engine [62]. It is claimed that the recirculation

area of the vapour phase is formed at the spark-plug position, providing an ignitable

mixture at the electrodes.

A great deal of research has been published for multihole nozzle sprays.

Figure 4.34 [95] shows a planar visualisation of the liquid and vapour concentra-

tion of a six-hole 50

◦

umbrella angle nozzle as a function of time after the start

of injection in a quiescent chamber using the laser-induced exciplex ﬂuorescence

technique. The features of relevance here are that fresh air is entrained at point A-1

284 Lean Flames in Practice

Figure 4.32. Piezoelectric injector: planar droplet–Mie scattering showing the liquid-fuel dis-

tribution at four time points from a single injection of 24-mg fuel. The injector is at the upper

left; a spark plug, superimposed on the images, is at the top centre. ASOI denotes after start

of injection [94].

as a result of the spray’s motion. At point B-2, a vortex caused by the propagating

spray transports vapour away from the liquid fuel: One can expect a similar, per-

haps more efﬁcacious, phenomenon to occur with an outward-opening piezoelectric

injector. As the injection event progresses, the spray at point C-2 starts to move

so as to effectively decrease the umbrella angle. By time D, when the injection has

ended, the resulting combustible vapour cloud resides within the umbrella angle

of the spray. Cycle-resolved simultaneous planar measurements of ﬂow ﬁeld and

quantitative equivalence ratio in a engine under motored and ﬁred conditions using

iso-octane fuel were made recently [96]. Note that measurements and analysis sug-

gest that the evaporation process is diffusion controlled and the effects of sequential

evaporation are expected to be insigniﬁcant on fuel stratiﬁcation [97]. This is part

of a research programme seeking correlations between marginal engine operation

and local instantaneous mixture and ﬂow conditions [98]. Figure 4.35 shows the cou-

pling of spray momentum with air–fuel mixing within a single cycle for a motored

4.1 Application of Lean Flames in Internal Combustion Engines 285

Figure 4.33. Piezoelectric injector: planar droplet–Mie scattering showing the liquid-fuel dis-

tribution at four time points from a split injection of 24 mg. ASOI denotes after start of

injection [94].

engine operating at stratiﬁed conditions with one of the sprays deliberately targeted

to strike the spark plug with the orientation shown: As might be expected from this

ﬁgure, spark-plug orientation can affect both the ignition delay and the 10%–50%

burn time [99]. Somewhat surprisingly, the majority of the liquid spray evaporated

within 2

◦

CA after the end of injection [100]. Also, the arc is strongly inﬂuenced by

the gas motion induced by the spray and its vapour: Broadly, conditions that lead to

shorter-duration arcs are unfavourable [101]. The ﬁrst part of the fuel-vapour cloud

has an equivalence ratio that is fuel rich but rapidly becomes fuel lean: This is also

showninFig.4.36, which is a scatter plot of spatial averages of equivalence ratio and

velocity magnitude from the region close to the spark plug shown in the inset.

Although the overall (‘delivered’) equivalence ratio to a stratiﬁed-charge en-

gine can be very lean, the optimum equivalence ratio at the time of ignition for

reliable ignition ranges from about 0.6 [87]to2[59]. These values represent a com-

promise between conditions at advance, which have much richer conditions and

high ﬂow velocities, and conditions at retard, which have lower and more favourable

286 Lean Flames in Practice

Upper laser boundary

Cone angle

Upper laser boundary

Cone angle

Upper laser boundary

Cone angle

Upper laser boundary

Figure 4.34. Multihole nozzle: direct imaging results using laser-induced exciplex ﬂuores-

cence from a 50

◦

‘umbrella’ multihole nozzle, showing qualitative time-resolved results from

measurements made in a chamber at 10 bars and 584 K. The results were processed so that the

vapour concentration is shown by the magnitude of the z axis, and the liquid concentration

is represented by the colour, red–high, blue–low in the z axis. ASOI is after start of injection

[95]. (See colour plate.)

velocities – say below 15 m/s to prevent spark stretching – but much leaner mix-

ture strength. The subsequent progress of combustion was gauged from spectrally

ﬁltered observation of the spatial and temporal development of emitted light. The

discussion in [87] describes this for a particular engine as a two-stage process, as

shown in Fig. 4.37, of the partially premixed combustion (visualised through OH*

chemiluminescence) followed by a second stage of mixing controlled combustion

of fuel-rich zones (visualised from the luminosity of hot soot, through black-body

emission wavelengths). For advanced ignition (SA = 34

◦

BTDC), when the equiv-

alence ratio is much richer at the spark plug, the results show both premixed and

rich combustion starting at the spark and propagating around the edges of the piston

bowl. Retarded ignition (SA = 22

◦

BTDC) results in much slower combustion than

for the early ignition.

Lean-Burn, Compression-Ignition Combustion

CONCEPTS. To retain the advantages of unthrottled operation and to burn the fuel

at yet leaner conditions than in DI SI engines, in the pursuit of further reduction

in NO

and smoke emissions, ideally to the point that no after-treatment is re-

quired at all, implies that the combustion in a premixed or partially premixed mode

should not exceed a temperature of about 1800 K. Temperatures below 1800 K avoid

the rapid rise in the generation of NO

and both any rich equivalence ratios and

4.1 Application of Lean Flames in Internal Combustion Engines 287

Figure 4.35. Simultaneous fuel concentration and velocity measurements inside a motored

spray-guided DI engine at stratiﬁed conditions with a multihole injector. One of the sprays

was deliberately aimed at the spark plug [96].

non-premixed ﬂames are to be avoided as they give rise to particulates. It is also

desirable to keep the temperature above about 1500 K to promote the conversion of

CO to CO

. These conditions imply operation near, or even below, the lean-mixture

limit, which is about φ = 0.5, depending on the amount of turbulence in the com-

bustion chamber and on the compression ratio [52]. Alternatively, mixture strengths

288 Lean Flames in Practice

Figure 4.36. Scatter plot of the spatial averages of fuel concentration and velocity magnitude

from 150 motored cycles. CAD denotes CA degrees [96].

up to stoichiometric can be used, but greatly diluted by ballast–bath gas: These are

not considered here. In principle such lean mixtures are ignited, in whole or in part,

through compression work and deliberately arranged to give rise to quasi-isochoric

combustion: This section outlines the practical consequences of adopting this strat-

egy. We note that this type of combustion in piston engines has many names in the

literature. Here we use two widely, but not universally, adopted terms, namely ho-

mogeneous charge compression ignition (HCCI) and low-temperature combustion

(LTC), the latter particularly for diesel-fuelled engines. The extent to which the for-

mer term, in particular, is accurate or appropriate has been debated and questioned

in the literature (see Subsection 3.1.3, and [102]), and there is no need to repeat that

debate here.

Although forms of HCCI have been known for some time, including some

designs of two-stroke engines for model aircraft that run on appropriate blends of

fuel, research on this topic began in the late 1970s and has since resulted in HCCI

combustion becoming commercially available, albeit in small numbers, in a couple of

two-stroke engines. In four-stroke diesel-fuelled engines, forms of LTC strategies re-

portedly found some commercial application since about 2000 [e.g., the ‘modulated

kinetics’ (MK) [103] and the ‘uniform bulky combustion’ (UNIBUS) [104] ap-

proaches]. For gasoline-fuelled engines, many manufacturers of four-stroke engines

have been researching this concept for some time [105, 106]. Other fuels, including

natural gas and dimethyl ether (DME), have also been investigated [107, 108].

4.1 Application of Lean Flames in Internal Combustion Engines 289

Figure 4.37. Effect of two values of ignition timing (denoted as SA = 22 and SA = 34, in

units of degrees CA BTDC) on the progress of combustion in a spray-guided stratiﬁed-charge

DI engine. Top row of each two-row set: visualisation through ensemble-averaged spectrally

resolved imaging: OH* chemiluminescence (blue) and radiative emission from soot (red and

green). Bottom row of each two-row set: CFD calculations of heat release rate (blue) and rich

combustion products (yellow) [87]. (See colour plate.)

Overview of HCCI Engines. In an HCCI engine, ideally all of the fuel is premixed

with the air before auto-ignition takes place, although the fuel mixture fraction (and

temperature) may well be – deliberately or otherwise – spatially inhomogeneous to

a greater or lesser extent. The premixing is achieved by use of a PFI or by injecting

directly into the combustion chamber: In the latter case, it may be useful for the

fuel injection equipment to be agile (in the sense that it can provide a considerable

degree of ﬂexibility in the number of injection pulses and the start and duration of

each pulse). Depending on the fuel and the compression ratio, compression work

provides some or all of the increase in the internal energy of the charge to promote

auto-ignition. If additional energy is required, it is provided by a variety of techniques,

including heating of the intake air [109], retention of residuals [110] or internal EGR

[111]. The start of ignition at a thermodynamically appropriate point in the cycle

implies some sort of control, usually through control of the initial temperature of the

reactants and of its subsequent development during compression. This is because

there is usually no ‘external’ event to promote ignition, such as a spark or fuel

injection as in conventional SI or CI engines (however, spark-assisted HCCI is an

area of active research [112–115] with conﬂicting reports on efﬁcacy [59]. The use of

a laser ignition source – based on the absorption of energy by moisture contained

290 Lean Flames in Practice

in the EGR at a wavelength of around 3 μm – was also recently investigated in a

vessel in [116]). Investigations into means of control of the phasing of combustion

over the range of speed and load for which HCCI/LTC can be achieved is thus one

of the main avenues of research.

Although auto-ignition of the premixed charge can be virtually instantaneous,

such isochoric combustion must be avoided. To avoid damage to the engine as the

charge burns, and to limit the noise, vibration, and harshness from the engine, the

rate of pressure rise must be limited to about 10 bars/

◦

CA. With increasing load, this

condition becomes increasingly hard to achieve and may eventually lead to the unde-

sirable phenomenon of engine knock at an equivalence ratio as low as about 0.3 [117].

Computational estimates suggest that, with a truly spatially homogeneous charge,

this rate of pressure rise would occur at a much lower equivalence ratio, and [118]

shows, on the basis of measurements of chemiluminescence, that ignition fortunately

starts at one or more isolated sites, rather than uniformly throughout the volume of

the combustion chamber, because of more-or-less inevitable inhomogeneities in lo-

cal conditions. This is corroborated by measurements of fuel distributions [119, 120].

It is argued that, in practice, it is the occurrence of either inevitable thermal strati-

ﬁcation or incomplete mixing between the fresh charge and hot residuals [121] that

gives rise to ‘sequential’ auto-ignition, starting with the hottest zone and proceeding

thus throughout the charge to the coolest. This reduces the rate of rise in pressure at

any given overall equivalence ratio. It is argued that it is mainly thermal stratiﬁcation

in the gases far from any thermal boundary layers that controls the maximum rate of

rise in pressure [118], and [117] concludes that means to enhance ‘naturally’ occur-

ring thermal stratiﬁcation remain a challenge. An interesting question is whether any

deﬂagration fronts coexist with the ‘volumetric’ combustion associated with auto-

ignition. Some hold that there is evidence for this, at least under some circumstances,

experimentally [122] and computationally [123]: Others hold that turbulence is likely

to extinguish ﬂames at such lean equivalence ratios. Recent debate is summarised in

[59] and is also mentioned in Subsection 3.1.3.

It is known that retarding combustion after TDC is effective in allowing oper-

ation at equivalence ratios in excess of 0.3, while respecting the limit to the rate

of pressure rise mentioned earlier by increasing the duration of burn. Dec and his

co-workers [124–126] argue, on the basis of single-zone and multizone chemical ki-

netic modelling, that this is due neither to the increase in the volume expansion

rate associated with retarding after TDC nor to the lowering of chemical reaction

rates associated with the lower combustion temperatures. Rather, it is due to an

‘ampliﬁcation’ of the effects of thermal stratiﬁcation: Expansion of the gases after

TDC increases the induction time. However, increasing retard leads [127] initially

to increased cycle-to-cycle variations, which are undesirable. These authors argue

that, for a given combustion phasing, the variations are smaller for a fuel exhibiting

two-stage, rather than single-stage, ignition (see subsequent discussion) by consid-

ering the higher rate of rise in temperature for a two-stage ignition fuel just before

so-called ‘hot ignition’. Further retarding leads to misﬁre and partial burning of the

charge: For a two-stage ignition fuel, this leads to complete loss of ignition in subse-

quent cycles, and, for a single-stage fuel, this is followed by an advanced cycle with

complete burning because of the presence of species in the partial burn cycle, which

enhances the auto-ignition of the following cycle. With an increase in the load, that

4.1 Application of Lean Flames in Internal Combustion Engines 291

100

Fuel Carbon Into Emissions &

Combust. Efficiency [%]

0.04 0.08 0.12 0.16

Equivalence Ratio [φ]

0.2 0.24 0.28 0.32

Comb. Eff.

Figure 4.38. Combustion efﬁciency and

emissions vs. equivalence ratio for well-

premixed iso-octane and air; 1200 rpm;

inj

= 135 kPa, CR = 14 [117].

is, with an increase in φ, combustion phasing must be between the knock and the

misﬁre limits. The angle between these decreases with an increase in φ and thus,

currently at least, HCCI is limited to loads that are about half that typical of SI or

CI operation. Thus it may be necessary to operate in two modes, with high power

being provided in gasoline-fuelled engines by either conventional SI stoichiometric

DI or, perhaps, by SI stratiﬁed DI. If this were the case, the maximum compression

ratio of the engine would be limited by the SI mode of operation.

The operation of the HCCI engine is also limited at low load – for example

operation at an equivalence ratio of, say, φ = 0.12 at idle [117] – by poor combustion

efﬁciency caused by high emissions of CO and UHCs. At these low loads, incomplete

bulk gas reactions begin to dominate [128] because the temperature is too low (at

higher-load operating points, these are from the near-wall regions and top land

crevice). Figure 4.38 [117] shows that this tendency starts at φ ≈ 0.2 for the particular

conditions of this ﬁgure. As the temperature falls below a level corresponding to a

peak charge temperature of 1500 K, the OH concentration decreases rapidly and

hence the main reaction for CO oxidation, namely CO + OH → CO

+ H, is

quenched by piston expansion. Figure 4.39 [129 ] suggests that fuel stratiﬁcation

by retarding injection can provide a solution but, at very retarded injection, NO

emission increases because regions of φ greater than about 0.6 arise.

100

Fuel Carbon Into Emissions &

Combust. Efficiency (%)

ISNO

(g/kWh)

0 45 90 135

Start of Injection (SOI) [°CA aTDC Intake]

180 225 315270 360

5.0

4.5

4.0

3.5

3.0

2.5

2.0

1.5

1.0

0.5

0.0

Combust. Eff.

Figure 4.39. Combustion efﬁciency and

emissions as functions of start of injec-

tion for an eight-hole injector. Low swirl

(SR = 0.9) and p

inj

= 120 bars [129].

292 Lean Flames in Practice

0.03

0.02

Mole Fraction

Temperature (K)

0.01

0.00

–5 0

Time after TDS [ms]

51015

1200

Temperature

neopentane

X 4

1100

1000

900

800

700

600

Figure 4.40. Computed temperature and

concentrations of neopentane, H

,and

OH, cited by [132].

Overview of Fuels, Including Single- and Two-Stage Auto-Ignition Characteristics. A

study of the chemistry of auto-ignition is fundamental to the understanding of the

operation of HCCI engines, and consequently much work has been published on

this aspect (e.g., [130–132]). Figure 4.40 [132] shows the particular example of a fuel

with two-stage ignition delay at a post-compression temperature of 757 K show-

ing proﬁles of temperature and concentrations of hydrogen peroxide and OH. The

ﬁrst stage (associated with ‘cool ﬂames’ and negative temperature coefﬁcients in

the variation of ignition delay times plotted against the reciprocal of temperature)

shows a delay of about 5 ms to a temperature of about 880 K with second-stage

ignition occurring at about 14 ms, when the temperature is between 900 and 1000 K.

There is a modest temperature increase associated with the ﬁrst ignition, which is the

result of low-temperature HC oxidation. Although a complete calculation requires

advanced reaction schemes, a readily appreciated description of the origin of two-

stage combustion is provided by [ 133]. This reference analyses two-stage ignition for

the particular example of n-heptane using largely an algebraic, rather than a com-

putational, approach based on two reduced four-step mechanisms. The ‘ﬁrst stage

ignition is related to a change from chain branching to chain breaking as a tempera-

ture threshold is crossed. The chain branching reactions build up ketohydroperoxides

which dissociate to produce OH radicals. As the temperature threshold is crossed,

this production ceases. The second stage is driven by the much slower production

of OH radicals owing to the dissociation of H

. The OH radicals react with the

fuel at nearly constant temperature until the latter is fully depleted’. A closed-form

solution elegantly describes the ﬁrst-stage ignition, the second-stage ignition, and

the origin of the crucial ‘temperature-crossover’ point, as well as its magnitude. The

crossover temperature is a weak function of pressure, at least over the range relevant

to automotive boosting.

It is of considerable practical importance in HCCI engines as to whether a

fuel exhibits a single- or two-stage ignition: The latter requires smaller amounts

of increase in temperature of the initial charge before compression work is added

or, equivalently, can operate at lower compression ratios. Typical fuels showing two-

stage ignition include diesel and n-heptane as well as blends of the primary reference

fuels with a sufﬁcient content of n-heptane [134]. Typical fuels showing single-stage

ignition under non-boosted conditions include gasoline, iso-octane, and ethanol. It

is the molecular structure of the fuel that controls the type of ignition that is ob-

served: Broadly [132], two-stage ignition is found in long, linear alkane fuel molecules