Fuel 144 (2015) 222–227

Contents lists available at ScienceDirect

Fuel

journal homepage: www.elsevier .com/locate / fuel

Analysis of pre-ignition to super-knock: Hotspot-induced deflagrationto detonation

http://dx.doi.org/10.1016/j.fuel.2014.12.0610016-2361/� 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author at: State Key Laboratory of Automotive Safety andEnergy, Tsinghua University, Beijing 100084, China. Tel.: +86 10 62772515; fax: +8610 62785708.

E-mail address: wangzhi@tsinghua.edu.cn (Z. Wang).

Zhi Wang a,⇑, Yunliang Qi a, Xin He b,a, Jianxin Wang a, Shijing Shuai a, Chung K. Law b,c

a State Key Laboratory of Automotive Safety and Energy, Tsinghua University, Beijing 100084, Chinab Center for Combustion Energy, Tsinghua University, Beijing 100084, Chinac Department of Mechanical and Aerospace Engineering, Princeton University, Princeton, NJ 08544, USA

h i g h l i g h t s

� Detonation in super-knock wasdemonstrated for the first time byoptical diagnostics.� Pre-ignition was captured under high

temperature and high pressureconditions.� Detonation is initiated in the

unburned mixture for closed system.� High pressure oscillation induced by

detonation.� Super-knock mechanism was

proposed as DDP.

g r a p h i c a l a b s t r a c t

a r t i c l e i n f o

Article history:Received 7 September 2014Received in revised form 9 December 2014Accepted 16 December 2014Available online 26 December 2014

Keywords:CombustionPre-ignitionSuper-knockDeflagrationDetonation

a b s t r a c t

Occurrence of sporadic super-knock is the main obstacle in the development of advanced gasolineengines. By utilizing a rapid compression machine, events of pre-ignition and super-knock in a closedsystem under high temperature and high pressure were captured by synchronous high-speed direct pho-tography and pressure measurement, with the results demonstrating that the mechanism of super-knockis constituted by hotspot-induced deflagration to detonation followed by high-pressure oscillation (DDP).

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

High boost and direct injection holds the potential of enhancedpower density and fuel consumption in the development of inter-

nal combustion engines (ICEs). Recent developments, especially inhighly boosted Spark Ignition (SI), gasoline-fueled engines withdirect injection in the low-speed and high-load operating regime,have however been challenged by the occurrence of a new modeof engine knock [1], which has been variously termed as super-knock, unwanted pre-ignition [2], mega knock [3], LSPI (low-speedpre-ignition) [4], Deto-knock [5], developing detonation [6,7] orsubsequent front propagation [8]. It is significant to note that sucha single super-knock event can instantaneously and severely

Z. Wang et al. / Fuel 144 (2015) 222–227 223

damage the engine due to the extremely high peak pressure andthe associated pressure oscillations developed. Furthermore,super-knock events appear sporadically with little direct relation-ship with engine control parameters such as ignition timing, air/fuel ratio and coolant temperature. Applying the common knocksuppression methods, such as retarding spark timing, enrichingmixture and enhancing wall heat transfer, are not effective atsuppressing the super-knock. As such, super-knock is at presentthe major obstacle for further improving the power density inturbo-charged SI engines.

Although numerous efforts to visualize the super-knock com-bustion process in ICEs have been made [9–14], no direct experi-mental observation has been reported that would allowquantitative analysis of a super-knock event, particularly there isno direct evidence of super-knock caused by detonation. This isbecause super-knock occurs at high loads, which is beyond theoperating conditions of conventional optical engines. This diffi-culty, however, can be circumvented by using a rapid compressionmachine (RCM) to simulate conditions similar to those within ICEs,while providing excellent optical accessibility. Indeed, by usingsuch a facility we have succeeded in observing, apparently forthe first time, the entire super-knock combustion process, fromdeflagration to detonation, under high temperature and high pres-sure conditions, as will be analyzed in this work.

2. Experimental setup

Fig. 1 shows the schematic of the RCM experimental setup.Briefly, the RCM has a diameter of 50.8 mm and is equipped witha piezoelectric pressure transducer for pressure measurement.The data were recorded at 100 kHz using a National Instrumentsdata acquisition system (cDAQ-9178 chassis coupled with analoginput model cDAQ-9223). The RCM is equipped with a quartz opti-cal window at the end wall, hence allowing visualization of theentire test section along its axial direction. Using a high-speedcamera (Photron Fastcam SAX2, Model 1000 K) with a Nikon50 mm lens (F1.4), color images were recorded at 45,000 frames

Fig. 1. The schematic of the R

per second with a CMOS array resolution of 512 ⁄ 512 pixels,resulting in an exposure time of 23 ls. Tests using a stoichiometriciso-octane/O2/Ar mixture were conducted under the operationconditions as listed in Table 1.

Fig. 2 shows the cross-section of the RCM combustion chamberwith the passages of intake, scavenging and pressure transducerand a representative combustion image from this view.

3. Results and discussion

3.1. Combustion visualization of pre-ignition to super-knock

Fig. 3 shows a representative pressure–time history in whichboth pre-ignition and super-knock were observed. Note that pre-ignition and super-knock were not observed in every RCM experi-ment. This is similar to the sporadic characteristics of super-knockin IC engines. Hundreds of RCM experiments were carried out andmost of them are homogeneous combustion. Only a few casesshowed pre-ignition with the pre-ignition sites randomly distrib-uted in the combustion chamber. For the case shown in Fig. 3,the pressure and temperature at the end of compression (time = 0)are 2.0 MPa and 932 K, respectively. The subsequent pressuredevelopment is largely similar to the super-knock trace recordedin SI engines [5], with three essential features occurring in thefollowing sequence: (1) gradual pressure rise, indicted by a secondpressure rise starting at 5.16 ms after compression; (2) strongpressure discontinuity at 6.99 ms; and (3) strong pressure oscilla-tion, with the amplitude of the maximum pressure rise being11.38 MPa. The knock intensity of this experiment, quantified bythe amplitude of the maximum pressure oscillation (Dp), is8.12 MPa. Since Dp of the maximum permitted knock intensity is0.5 MPa for SI engine in general, this is a typical super-knockaccording to the knock intensity (Dp > 5.0 MPa) defined in Ref.[5] based on experimental data statistics and combustion parame-ter calculations.

These distinguishing, common features clearly demonstrate thatthe combustion process of the RCM experiment is fundamentally

CM experimental setup.

Table 1Rapid compression machine specifications.

Bore (mm) 50.8Compression ratio 14.1Fuel IC8H18

Equivalence ratio 1Effective pressure (MPa) 20Effective temperature (K) 930Camera speed (fps) 45,000Ar/O2 3.76

-10 -5 0 5 10 15 200

2

4

6

8

10

12

6 7 8 90

4

8

12 enlarged view

pres

sure

(MP

a)

time (ms)

5 MPa

1.26MPa

6.99ms

pend=2.0MPa

Tend=932K

Fig. 3. Pressure trace of RCM showing deflagration, detonation and pressureoscillation.

224 Z. Wang et al. / Fuel 144 (2015) 222–227

similar to that of the super-knock in a boosted SI engine, which isclassified in the previous study [5].

Fig. 4 shows the time-stamped images of a high-speedsequence. Since the luminescence of the blue flame is very weakbefore 6.95 ms, the images for (a)–(i) were contrast-enhanced inorder to capture the flame propagation, as shown in the fourth row.

3.2. Analysis of the process from pre-ignition to super-knock

To facilitate interpretation of the observed super-knock event,Fig. 5 also shows selected images at specific instances over thepressure trace (starting at 5 ms after compression), which clearlydemonstrates the three distinguishing stages of the combustionprocess: deflagration propagation, detonation in unburned mix-ture, and pressure oscillation. They are discussed next.

3.2.1. 1st Stage: deflagration propagationFig. 5 shows that the combustion event is initiated by the

appearance of a random hotspot (white spot, it also can beobserved in Fig. 4a), representing a pre-ignition particle, that firstappears in the center of the test section at 5.16 ms and ignitesthe mixture. A deflagration wave is initiated and the resulting blueflame propagates outward. The radius of the flame reaches about19.25 mm in 1.79 ms, indicating an average flame front speed of10.7 m/s. The heat release from the deflagration leads to thermalexpansion of the burned mixture, which compresses the unburnedmixture to 3.2 MPa (1.26 MPa over the initial pressure of 2.0 MPa)and 1064 K.

We note in passing that while white spots were observed inother RCM experiments [15], and hypotheses proposed for theirorigin, no definite conclusion has been drawn.

3.2.2. 2nd Stage: detonation in unburned mixtureSubsequently at 6.97 ms, three distinct flames are initiated by

three corresponding explosion centers at the top-left corner nearthe cylinder wall, as shown in Fig. 5(k). These explosion centersare close to each other with two clearly identified merging

Fig. 2. The cross-section of RCM combustion chamber in c

boundaries. Since the mixture in the center of the test sectionhas already been burned, the newly-initiated flames wouldpreferentially propagate in the unburned mixture along theperiphery of the chamber, depicted in Fig. 5(k) as dotted yellowarrows. Subsequently, within 22 ls, the flame has propagatedthrough the majority of the unburned mixture with brightchemiluminescence emission at the periphery in the lower rightregion, as shown in Fig. 5(l). The small gap at the bottom, pointedby the red arrow in Fig. 5(l), indicates the front of the flame.

A rapidly propagating wave is subsequently visualized, which isfollowed by a reaction zone (transparent flame with blue chemillu-minescence emission) and travels from this region to the bottom-right corner via the surrounding unburned mixture. The passageoccurs in a short time, from 6.97 ms to 6.99 ms, yielding a propa-gation speed (vwall) of 3080 m/s along the yellow arrow in theunburned mixture near the wall. Since the calculated sound speedin the unburned mixture ahead of the flame is 558 m/s, theobserved wave is then propagating with a Mach number of 5.5,implying it is a supersonic, detonation wave. Indeed, theChapman–Jouguet theory yields a C–J detonation velocity (DCJ)of 2325 m/s (cp2 = 0.832 kJ/kg K, cp1 = 0.760 kJ/kg K, c2 = 1.364,q = 2223 kJ/kg, R2 = 0.222 kJ/kg K), demonstrating that the mea-sured wave speed, vwall, is not only of the same magnitude but isactually larger than DCJ. This clearly indicates that the bright blueflame in Fig. 5 from (k) to (l) is a strong detonation wave, whichrapidly sweeps through the pressure transducer, leading to a corre-spondingly rapid pressure rise from 3.26 MPa to 11.38 MPa in0.044 ms as is evident by the pressure ‘‘discontinuity’’ in Fig. 5.

amera view and a representative combustion image.

1) Natural luminosity images

2) Contrast-enhanced images

(

(

Fig. 4. High-speed images of combustion process in RCM (the number below each image represents the time in ms after the end of compression).

5.0 5.5 6.0 6.5 7.0 7.5 8.0 8.50

2

4

6

8

10

12

5.16ms

pre

ssur

e (M

Pa)

time (ms)

1.26MPa

6.99ms

t=0.09ms

pressure distinunity

1) Deflagration

2) Detonation

3) Pressure oscillation

a) e) j) k)

l)

m) n) o) w)

Pressuresensor

((((

(

((((

Fig. 5. Pressure trace with synchronous images showing three stages of super-knock process.

Z. Wang et al. / Fuel 144 (2015) 222–227 225

3.2.3. 3rd Stage: pressure oscillation in combustion chamberAfter initiation of the detonation, the propagating fronts collide

at the bottom of the combustion chamber, forming an incandes-cent zone in Fig. 5(m). The merged wave is then reflected at the

bottom of the wall, and bounces back towards the top of the cylin-der wall as shown in Fig. 5(m)–(o). A video showing the evolutionfrom deflagration, detonation to oscillation is given in the Supple-mentary Materials. Based on the locations of the wave front, the

-60 -40 -20 0 20 40 60 800

5

10

15

20

25

30

Cyl

inde

r pre

ssur

e (M

Pa)

Crank angle ( CA)

Spark timing

(end-gas autoignition)

(hot-spot induced detonation)

1390#: slight knock

1388#: super-knock

Normal cycle

954#: heavy knock(end-gas autoignition)

Fig. 7. Cylinder pressure traces of typical pre-ignition cycles and normal combus-tion cycle [5].

226 Z. Wang et al. / Fuel 144 (2015) 222–227

velocity of the resulting wave is calculated to be about 1155 m/s,which matches with the sound speed of the burned gas, indicatingthat it is now a compression wave since all the reactants in thecombustion chamber have been consumed at this stage. This resultis further collaborated by the recorded pressure oscillationdepicted in Fig. 5(k), which shows an oscillation frequency of11 kHz, representing a velocity of 1128 m/s in the radial direction.The entire combustion process is subsequently terminated, withthe amplitude of the pressure oscillation gradually attenuatesand the image becomes dark.

Detonation wave also induces gas motion. The detonation startson the top-left corner, which tends to push the gas moving fromtop-left towards bottom-right direction. The motion of the firstparticle that triggers the deflagration can be identified in theimages shortly after detonation. The flow velocity is about180 m/s calculated by tracking the location of the burning particlepointed by the green arrows shown in images (n) and (o).

3.3. Discussion of the combustion processes associated with the enginesuper-knock

High-speed imaging from the RCM experiment showed aknocking mechanism, which has not been previously reported.Here, a deflagration starts the combustion similar to thepre-ignition observed in IC engines. However, the deflagration doesnot self-accelerate to detonation as shown in Fig. 6. Instead, it isthe ‘‘hot spots’’ located in the unburned mixture that initiatesdetonation, which consumes the unburned mixture quickly andcauses large-magnitude pressure oscillation propagating at thelocal speed of sound. Similar combustion processes could takeplace in engines. As the start timing of ‘‘hot spot’’ in the unburnedmixture is crucial to detonation, this study helps to explain whyearlier pre-ignition timing does not always lead to higher knockintensity. If the ‘‘hot spot’’ starts too early, it may turn out to bea deflagration, similar to the combustion processes in the 1st stage.If the ‘‘hot spot’’ appears too late, the majority of the mixture couldhave already been consumed by the deflagration. As a result, thepressure rise will be smaller.

The above mechanism obtained by the RCM study can be usedto explain the phenomena of super-knock in gasoline engines.Fig. 7 shows the cylinder pressure traces of typical pre-ignitioncycles and normal combustion cycle in Ref. [5]. It is expected thatearly pre-ignition could lead to more severe engine knock. Unlikeregular knock, earlier pre-ignition timing does not always lead tohigher knock intensity. It is seen that, while cycle 954 has the ear-liest pre-ignition timing, the knock intensity is however actually

Fig. 6. Natural luminosity image (Fig. 5k in original resolution of 512 ⁄ 512 pixels)including pre-ignition, deflagration and detonation.

lower than cycle 1388 whose pre-ignition timing is after TDC butdetonation occurs in the end gas.

4. Concluding remarks

A combustion process associated with engine super-knock wasobserved in a rapid compression machine. Pre-ignition and super-knock processes were captured by synchronous high-speed directphotography and pressure measurement, demonstrating that thesuper-knock consists of three stages: (1) deflagration with a grad-ual pressure rise, (2) detonation with a strong pressure discontinu-ity, and (3) detonation followed with strong pressure oscillations.

A deflagration with slow flame front speed is initiated by a ‘‘hotspot’’. The deflagration does not self-accelerate to detonation.Instead, detonation is initiated by another ‘‘hot spot’’ located inthe unburned mixture with high temperature and high pressure,which causes rapid pressure rise and the follow-up pressure oscil-lation in the combustion chamber.

The present results not only show that strong detonation canoccur in the unburned mixture for closed system under high tem-perature and high pressure conditions, but also indicate that themechanism of super-knock may be described as DDP: hot-spotinduced deflagration to hot-spot induced detonation and followedby high pressure oscillation.

Acknowledgment

This study was supported by National Natural Science Founda-tion of China (Grant No. 51036004).

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.fuel.2014.12.061.

References

[1] Kalghatgi G. Development in internal combustion engines and implications forcombustion science and future transport fuels. In: Plenary lecture, 35thinternational combustion symposium, San Francisco, August 3–7, 2014; to bepublished in Proceedings of the Combustion Institute, vol. 35, 2014.

[2] Willand J, Daniel M, Montefrancesco E, Geringer B, Hofmann P, Kieberger M.Limits on downsizing in spark ignition engines due to pre-ignition. MTZWorldwide 2009;70:56–61.

[3] Winklhofer E, Hirsch A, Kapus P, Kortschak M, Philipp H. TC GDI engines at veryhigh power density? Irregular combustion and thermal risk, SAE technicalpaper 2009-24-0056.

[4] Amann M, Alger T, Mehta D. The effect of EGR on low-speed pre-ignition inboosted SI engines. SAE Int J Engines 2011;4:235–45.

Z. Wang et al. / Fuel 144 (2015) 222–227 227

[5] Wang Z, Liu H, Song T, Qi Y, He X, Shuai S, et al. Relationship between super-knock and pre-ignition. Int J Engine Res 2014. http://dx.doi.org/10.1177/1468087414530388.

[6] Bradley D, Morley C, Gu X, Emerson D. Amplified pressure waves duringautoignition: relevance to CAI engines. SAE technical paper 2002-01-2868.

[7] Bradley D, Kalghatgi G. Influence of autoignition delay time characteristics ofdifferent fuels on pressure waves and knock in reciprocating engines. CombustFlame 2009;156:2307–18.

[8] Bansal G, Im H. Autoignition and front propagation in low temperaturecombustion engine environments. Combust Flame 2011;158:2105–12.

[9] Dahnz C, Han K-M, Spicher U, Magar M, Schiessl R, Maas U. Investigations onpre-ignition in highly supercharged SI engines. SAE Int J Engines2010;3:214–24.

[10] Zahdeh A, Rothenberger P, Nguyen W, Anbarasu M, Schmuck-Soldan S,Schaefer J, et al. Fundamental approach to investigate pre-ignition inboosted SI engines. SAE Int J Engines 2011;4:246–73.

[11] Okada Y, Miyashita S, Izumi Y, Hayakawa Y. Study of low-speed pre-ignition inboosted spark ignition engine. SAE Int J Engines 2014;7:584–94.

[12] Dingle SF, Cairns A, Zhao H, Williams J, Williams O, Ali R. Lubricantinduced pre-ignition in an optical SI engine. SAE technical Paper 2014-01-1222.

[13] Döhler A, Pritze S. A contribution to a better understanding of pre-ignition inhighly supercharged direct-injection gasoline engines. In: 4th Internationalconference on knocking in gasoline engines, Berlin; 2013.

[14] Schünemann E, Witt A, Selder M, Schwarz C. Analyzing pre-ignition on asupercharged SIDI engine. In: 4th International conference on knocking ingasoline engines, Berlin; 2013.

[15] Walton SM, He X, Zigler BT, Wooldridge MS, Atreya A. An experimentalinvestigation of iso-octane ignition phenomena. Combust Flame 2007;150:246–62.