Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/254407873
CITATIONS READS
59 1,323
7 authors, including:
Mingzhi Song
China University of Mining Technology
11 PUBLICATIONS 74 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All in-text references underlined in blue are linked to publications on ResearchGate, Available from: Xudong Zhen
letting you access and read them immediately. Retrieved on: 20 November 2016
Applied Energy 92 (2012) 628–636
Applied Energy
journal homepage: www.elsevier.com/locate/apenergy
a r t i c l e i n f o a b s t r a c t
Article history: Knocking combustion in spark-ignition (SI) engines is an abnormal combustion phenomenon which can
Received 25 August 2011 constrain the engine performance and thermal efficiency. It can also result in severe permanent engine
Received in revised form 28 November 2011 damage under certain operating conditions. This paper systematically reviews the engine knock phenom-
Accepted 28 November 2011
enon, including the mechanisms, influencing factors, consequences and detection methods etc. It intro-
Available online 20 December 2011
duces the visualization researches, simulations and some judging indexes of engine knock. Three
mathematical models which can predict the engine knock, and various kinds of methods of suppressing
Keywords:
knock are summarized. Finally, this paper puts forward some new suggestions on the weakness in the
Knock
Detection methods
researches of knocking combustion.
Visualization researches Ó 2011 Elsevier Ltd. All rights reserved.
Simulations
Models
Suppressing methods
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 628
2. Knock detection methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 630
2.1. Methods based on in-cylinder pressure analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 630
2.2. Methods based on engine block vibration analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 630
2.3. Methods based on exhaust gas temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 630
2.4. Methods based on intermediate radicals and species analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 630
2.5. Methods based on heat release analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 631
3. Knock visualization researches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 631
4. Knock indexes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 631
4.1. The rate of heat release and cumulative heat release. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 632
4.2. MAPO, IMPG, IMPO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 632
4.3. Dimensionless Knock Indicator (DKI). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 632
4.4. Logarithmic Knock Intensity (LKI) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 632
5. Knock simulation researches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 632
6. Knock models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 633
6.1. AnB Knock model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 633
6.2. Shell Knock model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 633
6.3. Empirical formulations based on an Arrhenius function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 634
7. Methods of suppressing knock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 634
8. Conclusions and recommendations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 635
Acknowledgement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 635
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 635
0306-2619/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.apenergy.2011.11.079
X. Zhen et al. / Applied Energy 92 (2012) 628–636 629
[1]. Thus power generation with the lowest energy input from fos- of the unburned end-gas, so auto-ignition is seldom homogeneous,
sil resources has become more and more important since then. it usually occurs randomly in localized centers. When it occurs,
Due to the steady rise in oil prices and the increasing concerns pressure waves are generated, which can cause the formation of
towards global environmental pollution, the optimization of en- detonation waves, which sounds like metallic ringing which is dis-
gine performance and emission has become the most pressing fac- tinct from the mute sound of normal combustion. Sometimes auto-
tor. However, the optimization potential is not well utilized as ignition does not necessarily give rise to knock. There are three ba-
there is a lack of knowledge about the detailed in-cylinder combus- sic modes of propagation from the auto-ignition centers, depend-
tion process and emission formation mechanisms. Future stringent ing on the temperature gradients [16]. These are:
legislation on emissions in combination with an increase in engine
efficiency poses a great challenge to the designers and researchers. When the end-gas has low temperature and steep temperature
However, engine design and operating parameters such as com- gradients, it will produce a weak pressure, which propagates
pression ratio, inlet pressure and temperature, spark timing, equiv- from the center and is attenuated, combustion undergoes a
alence ratio and exhaust gas re-circulation must be optimized for gradual transition to knock. The cool flame, which travels with
the best performance and efficiency. The essential task in the average speeds of v = 50–200 m/s. In this phase, combustion is
development of SI engine today is to achieve a good balance be- non-knocking combustion.
tween the reduction of fuel consumption and the improvement When the end-gas has high temperature and small temperature
of torque [2,3]. gradients, it will generate simultaneous chemical reaction fol-
In SI engines, the combustion can proceed as a normal or abnor- lows the occurrence of the auto-ignition. As the onset of the
mal phenomenon depending on the certain operating conditions. main heat release, the average speed can be up to v = 500 m/s.
The normal one, which is initiated solely by a timed spark, gener- There is a clear correlation between the propagation velocity
ates a flame front moving across the cylinder volume in a uniform and the knock intensity: the faster the reaction front of main
manner at normal velocity. Depending mainly on the temperature heat release propagates, the higher the knock intensity.
and pressure history of the end-gas as well as on the rate of devel- When the end-gas has intermediate temperatures and temper-
opment of the flame, the abnormal combustion can cause an ature gradients. When the gradient smaller than a critical value,
important phenomenon: Knock. Knock is well known as a major it will generate enough intensity to initiate chemical reaction.
barrier obstructing further improvement of the SI engine thermal Then the strong shock waves are created. Pressure waves are
efficiency. If knock occurs over a long period of time, it will bring generated by the fast-propagating reaction front. When the
some unfavorable effects [4–6]: pressure is strong and the end-gas is reactive enough, it will
generate an intensely illuminating flame. The flame, which trav- Firstly, the sensors are very expensive. Besides, if the sensors
els opposite to the original temperature, its speeds up to are installed inside the cylinder chamber, it will direct contact
2000 m/s. In this phase, knocking combustion occurred. with hot and high pressure mixtures, so its lifetime expectancy
and accuracy will be reduced.
These three basic modes can represent limiting conditions and During non-knocking combustion conditions, the sensor mea-
are associated with trace, moderate, severe and damaging knock, sured the value at the sensor location can be representative of
respectively. Therefore, a complete understanding of knock can the whole combustion chamber pressure value, as knock occurs,
be of great benefit to respect the future emission normative as well huge non-homogeneities take place preventing the extension of
as to reduce development costs. the local data to the global domain. When knock occurs, it will
Describing and even predicting knock is complex, it requires generate pressure wave, but it evolves throughout a non-homo-
deep understanding on the processes occurring within the com- geneous medium, interacting with the complex chamber
bustion chamber. Many decades of exhaustive researches indicated boundary. So, the measured pressure wave amplitude has a
that the combustion mechanisms associated with knock were not local meaning, it cannot be considered as the maximum value
fully understood because of the complexity of the phenomena in- of the whole combustion chamber pressure wave.
volved [17]. There have been many experimental and simulation If one set for each cylinder, which not only increases cost, but
researches of the causes of engine knock. It would be beneficial also requires the use of various hardwares and softwares.
to develop a predictive technique for studying the causes of knock.
Knock prediction is a challenging task to designers and researchers Therefore, the method based on in-cylinder pressure analysis is
because of its complexity. extensively used in experimental researches.
2. Knock detection methods 2.2. Methods based on engine block vibration analysis
There are many methods to be used to detect knock, which can This method is to determine the occurrence of knock by mea-
be classified into two broad categories namely direct and indirect suring the vibration level of the engine. When knock occurs, it will
methods. The former one is based on the direct measurement create extensive pressure waves (above about 2 bar, peak to peak)
and study of inside cylinder parameters, which can be influenced within the combustion chamber. The pressure waves can radiate
by knock. Other methods are based on the indirect measurement sound within the audible frequency range directly, excite engine
such as sound pressure or cylinder block vibration [18–20]. block vibrations and produce the audible knock signal.
Knock detection can be performed based on several types of The frequencies of the oscillations highly depend on the size
methods, for instance, in-cylinder pressure analysis, cylinder block and shape of the chamber, the resonant modes being created and
vibration analysis, ion current analysis, light radiation and acoustic the local speed of sound, which in turn vary from one engine oper-
emissions analysis, heat transfer analysis, temperature analysis etc. ating point to another during combustion.
Here are some major application methods to review. Nonintrusive vibration sensors have a lot of advantages, for in-
stance, excellent durability and low-cost. Therefore, with the
2.1. Methods based on in-cylinder pressure analysis development of nonintrusive pressure sensors, the knock detection
methods based on block vibration analysis become the most prac-
This method is to analyze inside cylinder pressure signals to tical for mass-production car engines.
study directly the inside combustion processes, which are influ-
enced by knock. It is often used as a reference for the calibration
of detection strategies based on some sensors. The typical high- 2.3. Methods based on exhaust gas temperature
frequency knock signature is related to the combustion chamber
resonant frequencies, which are excited by the rapid pressure rise. Knock could be detected by monitoring exhaust gas temperature.
The typical pressure curves of a knocking and non-knocking cycle It is observed that there is clear correlation between engine knock
are shown in Fig. 1. and exhaust gas temperature. In knocking combustion conditions,
This method requires pressure signal processing. The expected exhaust gas temperature was shown to be reduced. This technique
resonant knock frequencies can first be estimated by using the ana- for detecting knock could be efficient for the following reasons [21]:
lytical solution of the wave equation. The resonant frequency of
the m, n vibration mode can be written as [20]: Because the signal will not be affected by engine noise, so the
knock detection can be pure.
fm;n ¼ am;n cs =ðpBÞ ð1Þ It is very quick and convenient.
It can be applied for all types of engines.
where am,n is the corresponding wave number (determined by
means of Bessel’s equations), m and n denote the numbers of radial
and circumferential pressures nodes respectively, cs is the speed of 2.4. Methods based on intermediate radicals and species analysis
sound inside the combustion chamber (estimated to be about
1000 m/s), B represents the cylinder bore. Nowadays, chemical luminescence emissions of end-gas can be
The definition of knock intensity usually uses the maximum used to analyze the chemical reactions caused by the end-gas auto-
amplitude of the filtered cylinder pressure, so the cylinder pressure ignition. Spectroscopic and chemiluminescence measurements
is band-pass filtered with a software filter. The lower cut-off fre- allow identifying CH, HCO, HCHO and OH radicals as markers of
quency is 4 kHz and the higher cut-off frequency is well above different combustion phases. For instance, HCHO is built up in
the frequency of the in-cylinder oscillations. the end gas before the flame front, hot spots can be detected by
The pressure wave amplitude can be used to define knock its higher concentration, CH and OH radicals can be markers of nor-
indexes: the drawback of such indexes is that they are based on mal combustion reaction and burned zone, respectively. Therefore,
observations referring to a narrow spatial domain. So the methods we can detect knock based on intermediate radicals and species
based on in-cylinder pressure analysis have some major drawbacks: analysis.
X. Zhen et al. / Applied Energy 92 (2012) 628–636 631
Fig. 2. Series of high-speed direct images and related in-cylinder pressure histories for both knocking and non-knocking engine cycles [24]. (1) Non-knocking engine cycle.
(2) Knocking engine cycle.
4. Knock indexes
4.1. The rate of heat release and cumulative heat release used to define the knock intensity of a running point. It should
be deeply researched.
When knock occurs, if neglect the heat exchanges, the heat
losses will increase and the net heat release should decrease. This 4.4. Logarithmic Knock Intensity (LKI)
approach is to look at the combustion characteristics in terms of
burning rate. The Rate of Heat Release (ROHR) is usually evaluated Many researches show that, as knock occurs, the useful pressure
based on the in-cylinder pressure signal, by applying the first law oscillation frequencies are between 6 and 25 kHz. When using
of thermodynamics [26]. higher sampling frequency, it has to handle more data, it requires
more memory and the computation time will be longer [30].
dV dP
ROHR ¼ ðc=ðc 1ÞÞP þ ð1=ðc 1ÞÞ ð2Þ The pressure oscillations are usually made up of several fre-
dh dh
quencies, an appropriate description of the knock intensity is the
The Cumulative Heat Release can be defined as: average energy. The equation of the average energy in terms of
Z the spectral data is [30]:
dV dP
CHRNET ¼ ðc=ðc 1ÞÞP þ ð1=ðc 1ÞÞV ð3Þ
dh dh X
25 kHz
Average Energy ¼ X 2 ðkÞ ð8Þ
where h is the crank angle degree, c is ratio of specific heats, P is cyl- 6 kHz
inder combustion pressure, V is the cylinder volume.
The equation of the average energy in the time domain is:
The abrupt decrease in the net cumulative heat release is likely
to be associated to a strong increase in the heat losses. The CHRNET 1XN1
parameter is strongly related to knock intensity, while it is not re- Average Energy ¼ x2 ðkÞ ð9Þ
N n¼0
lated to the sensor position.
where x(n) is the filtered pressure data (band pass filter 6–25 kHz),
4.2. MAPO, IMPG, IMPO and X(k) is the spectral data in 6–25 kHz of the original pressure sig-
nal, i.e.
MAPO, IMPG and IMPO, which are three knock indicators, based
1XN 1
on a high frequency analysis of cylinder pressure data, commonly XðkÞ ¼ wðnÞ xoriginal pressure ðnÞejð2p=NÞkn ð10Þ
N n¼0
used in the literature [27,28].
Maximum Amplitude of Pressure Oscillations (MAPOs) N is the number of samples used in both cases, w(n) is the Hamming
window function.
h þf
^jh0
MAPO ¼ max jp ð4Þ The knock intensity is finally defined as the logarithmic knock
0
N Z h0 þf
1 X Nowadays, computer simulation researches are more and more
IMPO ¼ ^jdh
jp ð6Þ
N 1 h0 widely used by many researchers. In SI engine knock researches,
more and more people also use computer simulations. Knocking
where N is the number of computed cycles, h0 is crank angle corre- simulation has a lot of methods including zero-dimensional
sponding to the beginning of the window of calculation, f is value of (0-D), one-dimensional (1-D), three-dimensional (3-D) etc. One-
the window of calculation, p ^ is the filtered in-cylinder pressure. dimensional models can model the entire engine cycle with
MAPO is related to the peak of the pressure oscillations due to acceptable computational times; however, they need semi-empir-
engine knock, IMPG is related to the modulus of pressure gradient ical correlations in order to model the flow field details and the
and IMPO represents the energy contained in the high frequency burning speed within each cylinder, so it cannot simulate the ac-
oscillations of the cylinder pressure signal (including noise). tual engine operating conditions, for instance, gas flow, in-cylinder
combustion and after-treatment etc. But 3-D simulation can solve
4.3. Dimensionless Knock Indicator (DKI) this problem. Recently, the development of integrated predictive
models for fuel auto-ignition remains as a challenge. Since the
DKI is a Dimensionless Knock Indicator built from existing knock propensity of fuels places an upper limit to engine perfor-
knock indices (IMPO and MAPO) [29], mance, a chemical-thermo fluid dynamic coupled model is a pow-
erful tool for optimizing engines and gasoline formulation. For
DKI ¼ IMPO=ðMAPO fÞ ð7Þ
instance, a transient knock prediction technique has been devel-
DKI is the ratio between two values that can be interpreted as oped by coupling a zero-dimensional knocking simulation with
two surfaces. Firstly, IMPO is the surface under the pressure signal chemical kinetics and a one-dimensional gas exchange engine
and secondly the quantity. MAPO f is the surface of the computa- model to study the occurrence of transient knock in SI engines [31].
tional window. DKI is an ‘‘image’’ of the knock intensity, which is For performance evaluation of a complicated energy system
calculated taking into account the ‘‘noise’’ before the auto-ignition. such as the internal combustion engine, Computational Fluid
The DKI value will decrease with the increase of knock intensity Dynamics (CFDs) techniques are more and more utilized in the
whatever the engine geometrical characteristics and settings are. development of new solutions for performance improvement of
Its use enables to detect the start of knock and then to determine engines. For instance CFD can significantly reduce the expensive
the KLSA (Knock Limited Spark Advance). and time consuming experiments necessary in the development
At present, DKI has not been widely applied, because it is ques- process of new engines or new engine parts. CFD techniques are
tionable that it could be considered better than others. Besides, able to provide detailed and sound information on engine phenom-
considering the cycle variation, more complex indexes have to be ena, but very often they are time consuming and hard to be imple-
X. Zhen et al. / Applied Energy 92 (2012) 628–636 633
mented [26,32–38]. CFD analysis can be used in order to simulate where a is a constant equal to 1 s1, Y T Fu is the mass fraction of fuel
knock effects on the in-cylinder pressure trace. The CFD models are tracer. The criteria of knocking combustion is satisfied when the
able to predict the combustion evolution with respect to spark ad- precursor concentration locally reaches the fuel tracer mass fraction
vance, from non-knocking to heavy knocking conditions. It can be value. Hence, when there is no combustion, the fuel mass fraction is
used to relate the actual combustion mode information to the mea- equal to the fuel tracer mass fraction. After auto-ignition, the
sured signal characteristics. CFD simulations can help defining change in precursor is simply calculated by:
knock indexes, the simulation results can be applied to assess the @yp
reliability of a methodology proposed. ¼ bY T Fu eð3500=T gb Þ ðq=qgf ÞðY Fu =Y T Fu Þ ð16Þ
@t
For these reasons, it is necessary to identify knock phenomena
by numerical calculation and by studying effective knock control- where b is a constant equal to 1000 s1, Tgb is the local temperature
ling methods. Knock modeling by means of 3D CFD simulation, of burnt gas, q is the density in the gas and qgf is the density in fresh
can be a powerful tool to understand and try to avoid this abnor- gas.
mal combustion phenomenon. We can use a three-dimensional The effect of air to fuel ratio on knocking tendency is accounted
model coupled with a one dimensional gas exchange model to pre- for by correcting the IO number. The octane number IOeff, corrected
dict transient knock phenomena of SI engines, and make a by the fuel/air ratio, is then:
3-dimensional CFD simulation coupled with detailed or reduced 2
with These are three most commonly used knock models, but there
x1 y1
are some differences in the use of approaches, for instance, both
f1 ¼ Af 1 expðEf 1 =RTÞ½O2 ½RH ; AnB and Shell model can be used with 0-D or 3-D approach, the
empirical approach described above can be used only in 0-D
f2 ¼ Af 2 expðEf 2 =RTÞ; approach.
It is found that the rapidity and severity of the onset of high phase is generally within the range of 6–25 kHz. In the evaluation
temperature ignition is probably associated with the transition of knock intensity, three knock indexes, such as MAPO, IMPG, IMPO
in these non-homogeneous regions to vigorous chain branching are adopted to evaluate the knock intensity. In the knock suppress-
through O atoms generation, the non-knocking reaction seems ing methods, delaying spark ignition time, using Split-injection and
to be restricted to conditions at which HO2 chemistry is able EGR technique etc. can effectively suppress the knock tendency.
to survive to late stages of the spontaneous ignition [56]. So fuel Despite great efforts made by researches, knocking combustion
design and blending may offer a route to a control of the knock. is still a crucial topic concerning engine design and development,
Use of cooled EGR (Exhaust Gas Re-circulation). It is well known and it needs further improvement. Generally speaking, higher
that residual gas has considerable suppressing effects on the flame temperature occurs closer to the flame front and it is more
auto-ignition tendencies of air/fuel mixtures. In SI engines, likely for auto-ignition to occur, but a number of experimental re-
EGR can be utilized to suppress the end-gas auto-ignition. How- searches have shown that auto-ignition usually occurs in the end-
ever, high temperature EGR may increase the intake charge gas area away from the flame front. However the current knock
temperature, and consequently increase the auto-ignition ten- theories cannot explain this phenomenon because: (1) The inter-
dency of the unburned end-gas. Therefore, normally the recircu- mediate radicals and species analysis among knocking combustion
lated exhaust gas needs to be cooled before being introduced phase requires further research, what kind of intermediate radicals
into the cylinder to achieve knock mitigation effect. Cooled and species are related to the occurrence, intensity and duration of
EGR has great potential in knock mitigation without the loss knocking combustion; (2) the detailed mechanisms of using EGR to
of output power [33,57–60]. Use direct fuel injection technol- suppress knock have also not been studied sufficiently, the influ-
ogy. Direct fuel injection can lower the combustion temperature ences of NOX species on the knock researches are not clear; (3)
due to the fuel vaporization cooling in the cylinder. Therefore it the current knock researches mainly focus on gasoline fuel, and
can lower knock sensitivity and the compression ratio can be the knock mechanisms of alternative fuels such as methanol, eth-
increased [50,61–65]. Use multiple injections. Split-injection is anol and natural gas have not been fully substantiated; (4) there
an effective method for mitigating the knock under high BMEP are some differences between experiments and simulation models,
operating conditions. However, the fuel-injection and combus- where the existing knock models should have made some certain
tion systems for a specific engine need to be carefully optimized assumptions and some key factors analysis such as the impact of
to obtain the merits of this strategy and to avoid performance gas flow may have been ignored. So the authors believe that future
deterioration. For instance, a ‘‘two-stage mixing’’ control strat- researches should focus on these areas, which not only could help
egy for knock suppression was proposed [61]. The first mixing explain the mechanisms of knocking combustion, but also have an
is to prepare the very lean premixed mixture, the second mixing important contribution to the development of internal combustion
is to prepare the distinctively stratified mixture. The premixed engine.
mixture is too lean to lead to knock, the stratified mixture does
not have enough time to proceed the reaction for knock. Use
stratified stoichiometric mixture. Stratified stoichiometric mix- Acknowledgement
ture can suppress knocking and knocking intensity is decreasing
with central zone richening. It is demonstrated that stratified This work was supported by the National Natural Science Foun-
stoichiometric mixture by two-stage injection strategy in GDI dation of China (Grant No. 51176137).
engine can suppress knocking effectively [66,67].
It is found that the occurrence of auto-ignition in the end gas is
sensitive on the amount of nitric oxide in the rest gas of the SI References
engine. The presence of NO in the initial gas has a strong influ-
ence on knock onset [68]. So knock can be suppressed by con- [1] Gan S, Ng HK, Pang KM. Homogeneous charge compression ignition (HCCI)
combustion: implementation and effects on pollutants in direct injection
trolling the concentrations of NO in the rest gas. diesel engines. Appl Energy 2011;88:559–67.
The use of direct injection (DI) of a second fuel, ethanol or meth- [2] Stone R. Introduction to internal combustion engines. 3rd ed. London:
anol (or their concentrated blends), is a means of avoiding Macmillan Press Ltd.; 1999.
[3] Nakama K, Kusaka J, Daisho Y. Study of knock control in small gasoline engines
knock in spark-ignited engines. For instance, the dynamics of
by multi-dimensional simulation. SAE Technical Paper no. 2006-32-0034;
alcohol vaporization has substantial impact on the temperature 2006.
of the unburned gas in which knock occurs. The concept of [4] Nates RJ, Yates ADB. Knock damage mechanisms in spark-ignition engines. SAE
direct alcohol injection for knock control was an effective Technical Paper no. 942064; 1994.
[5] Fitton J, Nates R. Knock erosion in spark-ignition engines. SAE Technical Paper
method [69,70]. no. 962102; 1996.
[6] Konig G, Maly RR, Bradley D, Lau AKC. Sheppard CGW. Role of exothermic
8. Conclusions and recommendations centres on knock initiation and knock damage. SAE Technical Paper no.
902136; 1990.
[7] Warnatz J, Maas U, Dibble RW. Combustion: physical and chemical
This paper systematically reviews the engine knock phenome- fundamentals, modeling and simulation, experiments, pollutant formation.
non, which includes the mechanisms, influencing factors and 4th ed. Berlin, Heidelberg, New York: Springer; 2006.
[8] Konig G, Sheppard CGW. End gas auto-ignition and knock in a spark ignition
detection methods etc. The visual experiment and simulation re- engine. SAE Technical Paper no. 902135; 1990.
searches have shown that knock phenomenon is the process of [9] Hirst SL, Kirsch LJ. The application of a hydrocarbon auto-ignition model in
one or more auto-ignition kernels formation and rapid propagation simulating knock and other engine combustion phenomena, in: Combustion
modeling in reciprocating engines. New York: Plenum; 1980.
with the speed up to 2000 m/s. In the knock prediction researches, [10] Westbrook CK, Pitz WJ, Leppard WR. The autoi-gnition chemistry of paraffinic
there are three most commonly used knock models, namely, AnB fuels and pro-knock and anti-knock additives: a detailed chemical kinetic
Knock model, Shell Knock model and Empirical formulations based study. SAE Technical Paper no. 912314; 1991.
[11] Vermeersch ML, Held TJ, Stein Y, Dryer FL. Auto-ignition chemistry studies of
on an Arrhenius function. There is a clear relation between the
n-butane in a variable pressure flow reactor. SAE Technical Paper no. 912316;
propagation velocity and the knock intensity: the faster the reac- 1991.
tion front of main heat release propagates, the higher the knock [12] Cowart JS, Haghgooie M, Newman CE, Davis GC, Pitz WJ, Westbrook CK. The
intensity and the tendency to the formation of knock as such be- intensity of knock in an internal combustion engine: an experimental and
modeling study. SAE Technical Paper no. 922327; 1992.
come. The spectrum analysis of knocking combustion phase has [13] Moses E, Yarin AL, Yoseph PB. On knocking prediction in spark ignition
shown that resonance frequency during knocking combustion engines. Combust Flame 1995;101:239–61.
636 X. Zhen et al. / Applied Energy 92 (2012) 628–636
[14] Rothe M, Heidenreich, Spicher U, Schubert A. Knock behavior of SI-Engines: [43] Halstead MP, Kirsch LJ, Quinn CP. The autoignition of hydrocarbon fuels at high
thermodynamic analysis of knock onset locations and knock intensities. SAE temperatures and pressures-fitting of a mathematical model. Combust Flame
Technical Paper no. 2006-01-0225; 2006. 1977;30:45–60.
[15] Kawahara N, Tomita E. Visualization of auto-ignition and pressure wave [44] Livengood JC, Wu PC. Correlation of autoignition phenomena in internal
during knocking in a hydrogen spark-ignition engine. Int J Hydrogen Energy combustion engines and rapid compression machines. In: 5th Symposium
2009;34:3156–63. (International) on combustion. vol. 347. 1955.
[16] Poschl M, Sattelmayer T. Influence of temperature inhomogeneities on [45] Towers JM, Hoekstra RL. Engine knock, a renewed concern in motorsports – a
knocking combustion. Combust Flame 2008;153:562–73. literature review. SAE Technical Paper no. 983026; 1998.
[17] Radu B, Martin G, Chiriac R, Apostolescu N. On the knock characteristics of LPG [46] Russ S. A review of the effect of engine operating conditions on borderline
in a spark ignition engine. SAE Technical Paper no. 2005-01-3773; 2005. knock. SAE Technical Paper no. 960497; 1996.
[18] Worret R, Bernhardt S, Schwarz F, Spicher U. Application of different cylinder [47] Rahmounia C, Brecq G, Tazerout M, Corre OL. Knock rating of gaseous fuels in a
pressure based knock detection methods in spark ignition engines. SAE single cylinder spark ignition engine. Fuel 2004;83:327–36.
Technical Paper no. 2002-01-1668; 2002. [48] Saikaly K, Corre OL, Rahmouni C, Truffet L. Preventive knock protection
[19] Corti E, Forte C. Statistical analysis of indicating parameters for knock technique for stationary SI engines fuelled by natural gas. Fuel Process Technol
detection purposes. SAE Technical Paper no. 2009-01-0237; 2009. 2010;91:641–52.
[20] Millo F, Ferraro CV. Knock in S.I. engines: a comparison between different [49] Li HL, Karim GA. Knock in spark ignition hydrogen engines. Int J Hydrogen
techniques for detection and control. SAE Technical Paper no. 982477; 1998. Energy 2004;29:859–65.
[21] Mohamad AQ. Exhaust gas temperature for knock detection and control in [50] Mohammadi A, Shioji M, Nakai Y, Ishikura W, Tabo E. Performance and
spark ignition engine. Energy Convers Manage 1996;37:1383–92. combustion characteristics of a direct injection SI hydrogen engine. Int J
[22] Grandin B, Denbratt I. The effect of knock on heat transfer in SI engines. SAE Hydrogen Energy 2007;32:296–304.
Technical Paper no. 2002-01-0238; 2002. [51] Celik MB, Ozdalyan B, Alkan F. The use of pure methanol as fuel at high
[23] Ollivier E, Bellettre J, Tazerout M, Roy GC. Detection of knock occurrence in a compression ratio in a single cylinder gasoline engine. Fuel 2011;90:1591–8.
gas SI engine from a heat transfer analysis. Energy Convers Manage [52] Selim MYE. Sensitivity of dual fuel engine combustion and knocking limits to
2006;47:879–93. gaseous fuel composition. Energy Convers Manage 2004;45:411–25.
[24] Kawahara N, Tomita E, Sakata Y. Auto-ignited kernels during knocking [53] Lanzafame R. Water injection effects in a single-cylinder CFR Engine. SAE
combustion in a spark-ignition engine. P Combust Inst 2007;31:2999–3006. Technical Paper no. 1999-01-0568; 1999.
[25] SchieBl R, Maas U. Analysis of endgas temperature fluctuations in an SI engine [54] Waters PF. A physical solution to engine knock, SAE Technical Paper no. 2002-
by laser-induced fluorescence. Combust Flame 2003;133:19–27. 01-2835; 2002.
[26] Forte C, Corti E, Bianchi GM. Combined experimental and numerical analysis of [55] Brecq G, Bellettre J, Tazerout M, Muller T. Knock prevention of CHP engines by
knock in spark ignition engines. In: Proceedings of the ASME 2009 internal addition of N2 and CO2 to the natural gas fuel. Appl Therm Eng
combustion engine division fall technical conference. ICEF2009-14102; 2009. 2003;23:1359–71.
[27] Cavina N, Corti E, Minelli G, Moro D, Solieri L. Knock indexes normalization [56] Griffiths JF, Whitaker BJ. Thermokinetic interactions leading to knock during
methodologies. SAE Technical Paper no. 2006-01-2998; 2006. homogeneous charge compression ignition. Combust Flame 2002;131:386–99.
[28] Brecq G, Corre OL. Modeling of in-cylinder pressure oscillations under [57] Grandin B, Angstrom HE, Stalhammar P, Olofsson E. Knock suppression in a
knocking conditions: introduction to pressure envelope curve. SAE Technical turbocharged SI engine by using cooled EGR. SAE Technical Paper no. 982476;
Paper no. 2005-01-1126; 2005. 1998.
[29] Brecq G, Bellettre J, Tazerout M. A new indicator for knock detection in gas SI [58] Grandin B, Angstrom H. Replacing fuel enrichment in a turbo charged SI
engines. Int J Therm Sci 2003;42:523–32. engine: lean burn or cooled EGR. SAE Technical Paper no. 1999-01-3505; 1999.
[30] Hudson C, Gao X, Stone R. Knock measurement for fuel evaluation in spark [59] Liang L. Multidimensional modeling of combustion and knock in spark-
ignition engines. Fuel 2001;80:395–407. ignition engines with detailed chemical kinetics. Ph.D. Thesis, University of
[31] Noda T, Hasegawa K, Kubo M, Itoh T. Development of transient knock Wisconsin-Madison; 2006.
prediction technique by using a zero-dimensional knocking simulation with [60] Fontana G, Galloni E. Experimental analysis of a spark-ignition engine using
chemical kinetics. SAE Technical Paper no. 2004-01-0618; 2004. exhaust gas recycle at WOT operation. Appl Energy 2010;87:2187–93.
[32] Colin O, Benkenida A, Angelberger C. 3D modeling of mixing, ignition and [61] Kuwahara K, Ueda K, Ando H. Mixing control strategy for engine performance
combustion phenomena in highly stratified gasoline engines. Oil Gas Sci improvement in a gasoline direct injection engine. SAE Technical Paper no.
Technol 2003;58:47–62. 980158; 1998.
[33] Liang L, Reitz RD, Iyer CO, Yi JW. Modeling knock in spark-ignition engines [62] Kume T, Iwamoto Y, Iida K, Murakami M, Akishino K, Ando H. Combustion
using a G-equation combustion model incorporating detailed chemical control technologies for direct-Injection SI engine. SAE Technical Paper no.
kinetics. SAE Technical Paper no. 2007-01-0165; 2007. 960600; 1996.
[34] Fontana G, Galloni E, Palmaccio R, Torella E. Numerical analysis of a small [63] Topfer G, Reissing J, Weimar HJ, Spicher U. Optical Investigation of knocking
turbo-charged spark-ignition engine. In: Proceedings of the ASME internal location on SI-engines with direct-injection. SAE Technical Paper no. 2000-01-
combustion engine division 2006 spring technical conference. ICES2006-1336; 0252; 2000.
2006. [64] Suzuki Y, Oki H, Kurosawa S. Gasoline S.I. engine combustion and knocking
[35] Bozza F, Fontana G, Galloni E, Torella E. 3D-1D analyses of the turbulent flow analysis on the mixture formation in the various direct injection fuel spray
field, burning speed and knock occurrence in a turbocharged SI engine. SAE patterns. SAE Technical Paper no. 2006-01-0230; 2006.
Technical Paper no. 2007-24-0029; 2007. [65] Schwarz C, Schünemann E, Durst B, Fischer J, Witt A. Potentials of the spray-
[36] Elmqvist C, Lindstrom F, Angstrom HE, Grandin B, Kalghatgi G. Optimizing guided BMW DI combustion system. SAE Technical Paper no. 2006-01-1265;
engine concepts by using a simple model for knock prediction. SAE Technical 2006.
Paper no. 2003-01-3123; 2003. [66] Bai YL, Wang Z, Wang JX. Simulation research on knocking suppressing using
[37] Ibrahim A, Bari S. Optimization of a natural gas SI engine employing EGR stratified stoichiometric mixture in a GDI engine. Trans CSICE 2010;28:393–8.
strategy using a two-zone combustion model. Fuel 2008;87:1824–34. [67] Bai YL, Wang Z, Wang JX. Experimental research on knocking suppression
[38] Kleemann AP, Menegazzi P, Henriot S, Marchal A. Numerical study on knock using stratified stoichiometric mixture in a GDI engine. Trans CSICE
for an SI engine by thermally coupling combustion chamber and cooling 2009;27:534–9.
circuit simulations. SAE Technical Paper no. 2003-01-0563; 2003. [68] Stenlaas O, Gogan A, Egnell R, Sunden B, Mauss F. The influence of nitric oxide
[39] Lafossas FA, Castagne M, Dumas JP, Henriot S. Development and validation of a on the occurrence of auto-ignition in the end gas of spark ignition engines. SAE
knock model in spark ignition engines using a CFD code. SAE Technical Paper Technical Paper no. 2002-01-2699; 2002.
no. 2002-01-2701; 2002. [69] Blumberg PN, Bromberg L, Kang H, Tai C. Simulation of high efficiency heavy
[40] Sazhin SS, Sazhina EM, Heikal MR, Marooney CJ, Mikhalovsky SV. The Shell duty SI engines using direct injection of alcohol for knock avoidance. SAE
auto-ignition model: a new mathematical formulation. Combust Flame Technical Paper no. 2008-01-2447; 2008.
1999;177:529–40. [70] Bromberg L, Cohn DR, Heywood JB. Calculations of knock suppression in highly
[41] Sazhina EM, Sazhin SS, Heikal MR, Marooney CJ. The Shell autoignition model: turbocharged gasoline/ethanol engines using direct ethanol injection. MIT
applications to gasoline and diesel fuels. Fuel 1999;78:389–401. Laboratory for Energy and the Environment report LFEE 2006-001; April 2005.
[42] Halstead MP, Kirsch LJ, Prothero A, Quinn CP. A mathematical model for
hydrocarbon autoignition at high pressures. Proc R Soc London
1975;346:515–38.