(KAIST)] Hyemin Kim Machine, Seung Wook Baek & Daejun ...procom.kaist.ac.kr/Download/IJP/136.pdf · (KAIST)] On: 30 May 2014, At: 19:10 Publisher: Taylor & Francis Informa Ltd Registered

This article was downloaded by: [Korea Advanced Institute of Science & Technology(KAIST)]On: 30 May 2014, At: 19:10Publisher: Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Combustion Science and TechnologyPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/gcst20

Auto-Ignition Characteristics of Single n-Heptane Droplet in a Rapid CompressionMachineHyemin Kima, Seung Wook Baeka & Daejun Changb

a Division of Aerospace Engineering, Korea Advanced Institute ofScience and Technology, Daejeon, Koreab Division of Ocean System Engineering, Korea Advanced Institute ofScience and Technology, Daejeon, KoreaAccepted author version posted online: 18 Feb 2014.Publishedonline: 18 Feb 2014.

To cite this article: Hyemin Kim, Seung Wook Baek & Daejun Chang (2014) Auto-IgnitionCharacteristics of Single n-Heptane Droplet in a Rapid Compression Machine, Combustion Science andTechnology, 186:7, 912-927, DOI: 10.1080/00102202.2014.890598

To link to this article: http://dx.doi.org/10.1080/00102202.2014.890598

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the“Content”) contained in the publications on our platform. However, Taylor & Francis,our agents, and our licensors make no representations or warranties whatsoever as tothe accuracy, completeness, or suitability for any purpose of the Content. Any opinionsand views expressed in this publication are the opinions and views of the authors,and are not the views of or endorsed by Taylor & Francis. The accuracy of the Contentshould not be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilities whatsoever orhowsoever caused arising directly or indirectly in connection with, in relation to or arisingout of the use of the Content.

This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &

Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Dow

nloa

ded

by [

Kor

ea A

dvan

ced

Inst

itute

of

Scie

nce

& T

echn

olog

y (K

AIS

T)]

at 1

9:10

30

May

201

4

Combust. Sci. Technol., 186: 912–927, 2014Copyright © Taylor & Francis Group, LLCISSN: 0010-2202 print / 1563-521X onlineDOI: 10.1080/00102202.2014.890598

AUTO-IGNITION CHARACTERISTICS OF SINGLEn-HEPTANE DROPLET IN A RAPID COMPRESSIONMACHINE

Hyemin Kim,1 Seung Wook Baek,1 and Daejun Chang2

1Division of Aerospace Engineering, Korea Advanced Institute of Scienceand Technology, Daejeon, Korea2Division of Ocean System Engineering, Korea Advanced Institute of Scienceand Technology, Daejeon, Korea

The autoignition characteristics of single n-heptane droplets inside a rapid compressionmachine (RCM) were investigated. During the compression stroke, the temperature andpressure inside the reaction chamber both rise rapidly, and subsequently decrease after thepiston reaches top dead center. When a fuel droplet experiences these transient conditions,if sufficient vaporization occurs, the droplet can autoignite. A single n-heptane droplet wasplaced at the center of the reaction chamber. The droplet was suspended from the tip of a 50-µm-diameter thermocouple, and its transient bulk temperature was measured. The evolutionof the droplet was recorded using a high-speed charge-coupled device array camera with aframe rate of 500 fps. The initial droplet diameter was in the range of 400−1000 µm, and thecompression rate was varied; compression stroke durations of 185 ms and 235 ms were inves-tigated. The ignition delay was longer when the initial droplet diameter was larger due to theslower heating of the droplet. Ignition did not occur if the droplet exceeded a critical diam-eter because sufficient fuel vapor was not generated during compression. The ignition delaywas shorter for the 185-ms compression stroke owing to the droplet evaporation dynamics.

Keywords: Autoignition; Droplet; Rapid compression machine; Transient condition

INTRODUCTION

The ignition characteristic of hydrocarbon fuel droplets is an important parameterin the performance of combustion devices, such as internal combustion engines, turbines,and rocket engines (Roth and Heidelberg, 2000). In these systems, fuel is injected intoa combustor in a spray, thereby generating the atomized droplets through the combustor.Evaporation of fuel droplets occurs when they are exposed to the high-temperature environ-ment inside the combustor. Autoignition of droplets can occur if the concentrations of fuelvapor and oxidizer, and the temperature in the vicinity of the droplets are sufficiently high.In general, the time between the exposure of droplets to the high-pressure and temperature

Received 2 September 2013; revised 23 December 2013; accepted 30 January 2014.Address correspondence to Dr. Seung Wook Baek and Dr. Daejun Chang, Engineering Bldg., Korea

Advanced Institute of Science and Technology (KAIST), 373-1 Guseong-dong, Yuseong-gu, Daejeon 305-701,Korea. E-mail: [email protected]; [email protected]

Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/gcst.

912

Dow

nloa

ded

by [

Kor

ea A

dvan

ced

Inst

itute

of

Scie

nce

& T

echn

olog

y (K

AIS

T)]

at 1

9:10

30

May

201

4

AUTO-IGNITION OF n-HEPTANE DROPLET 913

environment of the combustor and the onset of ignition is critically important in the analysisof combustion phenomena, and is termed the ignition delay. The autoignition characteristicsof droplets are affected by several parameters, including the type of fuel, the surroundinggas temperature, and the chemical kinetics of the fuel vapor. Moreover, the heat and massdiffusion processes in the vicinity of the droplets are also significant for ignition (Borghesiet al., 2011). It is, therefore, very important to understand the autoignition behavior ofdroplets to predict ignition phenomena inside a combustion system.

Several detailed studies of droplet ignition phenomena have been reported. Khanet al. (2007) carried out an experimental study of the autoignition and combustion behaviorof kerosene droplets at elevated temperatures and pressures. They found that the ignitiondelay exhibited Arrhenius temperature variation over the entire experimental range. In addi-tion, they reported an empirical relationship describing the ignition delay of various fueldroplets. Yang and Wong (2003) conducted a numerical simulation of the ignition of n-heptane droplets in a zero-gravity environment under high temperature and pressure. Theycompared their simulated data with measured data and found good agreement betweendatasets. Stauch et al. (2006) simulated the effects of the temperature and pressure of thesurrounding air conditions on the autoignition of a single droplet. The location of the igni-tion was investigated for each set of ignition conditions, and a detailed model describingthe chemical reaction mechanisms was provided. Wang and Baek (2007) simulated theunsteady behavior of a spray when fuel was injected into a pressurized combustor. Theyfound that the significant effects of the interaction between the droplets on the ignition timedepended on the pressure inside the combustor.

In most of previous reports, the temperature and pressure were stable, both temporallyand spatially. Indeed, it is difficult to account for transient conditions due to the complexityof the problem. However, in practical devices, the temperature and pressure, that dropletsare exposed to, vary rapidly with time. For example, when fuel droplets are injected dur-ing the compression stroke in a diesel homogeneous charge compression ignition (HCCI)engine, the in-cylinder temperature and pressure vary rapidly (Ma et al., 2008). In such asituation, the evaporation and ignition behavior of the fuel droplets are expected to differsignificantly from those of the steady state.

There are a number of reports of investigations of the in-cylinder ignition of spraydroplets. A rapid compression machine (RCM) (Akiyama et al., 1998) or rapid compressionexpansion machine (RCEM) (Hong et al., 1991) can be used to simulate conditions insidean internal combustion engine, and bulk combustion phenomena inside the cylinder havebeen observed. Nonetheless, such experiments are not suitable for observing the behaviorof a single droplet because the effects of neighboring droplets are significant. In this study,a single fuel droplet was placed in an RCM to eliminate the effects of interaction betweenneighboring droplets and the effects of combustion products. Thus, we sought to observethe fundamental behavior of a single fuel droplet during ignition under rapidly time-varyingconditions.

The primary goal of this study is to experimentally observe the ignition character-istics of a single droplet in a rapidly time-varying environment using an RCM. As thevolume of the reaction chamber decreases, the pressure and temperature inside the cham-ber increase, and then subsequently decrease once the piston has reached top dead center(TDC) due to heat loss through the walls of the machine. A single fuel droplet was sus-pended at the tip of a 50-µm-diameter fine thermocouple, and the temperature of the dropletwas measured. The ignition delay was measured by imaging the droplet using a high-speedcamera. Moreover, we used numerical models to simulate the conditions inside the RCM,

Dow

nloa

ded

by [

Kor

ea A

dvan

ced

Inst

itute

of

Scie

nce

& T

echn

olog

y (K

AIS

T)]

at 1

9:10

30

May

201

4

914 H. KIM ET AL.

and compared the results with our measured data. These simulations were used to explainthe transient behavior of the droplets. This approach can be applied to develop ignitioncontrol strategies for direct injection (DI)-HCCI engines, which is currently an importantchallenge (Ma et al., 2008).

EXPERIMENTAL APPARATUS

A schematic diagram of the experimental apparatus is shown in Figure 1a. Theexperimental system comprised an RCM, sensors, and an optical system that includeda high-speed charge-coupled device (CCD) array camera and a post-processing system.The time-synchronized data, including the temperature, pressure, and droplet images, werelogged using a personal computer (PC).

RCM

An RCM is a device for simulating homogeneous charge compression conditionsusing a single compression stroke. Reports describing the use of RCMs (Guibert et al.,2010; Kukkadapu et al., 2012) cite advantages that include the elimination of combustionproducts and the availability of optical observation. The RCM used in the experiment isshown in Figure 1b. It consisted of a reaction chamber, a driving chamber, and a drivingpiston. The bore of the driving piston was 130 mm, and the diameter of the reaction cham-ber was 50 mm. A crevice design was used for the reaction chamber piston to prevent theformation of corner vortices (Lee and Hochgreb, 1998; Mittal and Sung, 2006). Detaileddimensions of the crevice piston are shown in Figure 1b. The piston was operated pneu-matically, and the compression rate was controlled by varying the pressure in the range10–40 bar. The reaction chamber could withstand internal pressures in excess of 200 bar.Optical access was provided through 10-mm-diameter quartz windows installed in the reac-tion chamber. Pressure sensors and a thermocouple were installed in the wall of the RCM.The clearance between the driving chamber and the driving piston allowed the impact to beabsorbed, which prevented damage to the piston (Mittal and Sung, 2006).

Thermocouple and Pressure Transducer

Several methods have been proposed for placing a single droplet inside a combus-tor, including use of a thermocouple (Watanabe et al., 2010), quartz fiber (Ghassemi et al.,2006), or free-falling droplet (Honnery et al., 2013). Among these, the thermocouple hasseveral advantages, including stable and repeatable droplet placement and, importantly,straightforward measurement of the droplet temperature. In this study, the droplet was sus-pended from the tip of a Ch−Al sheathed, K-type thermocouple (Omega Engineering, Inc.).The cover tip of the thermocouple was removed, and a bead was formed by welding thetwo inner wires. The diameter of the two fused inner wires was 50 µm, and the diameterof the bead was <100 µm. The diameter of the droplets was on the order of a few hundredmicrons, and each droplet was placed at the tip of the thermocouple by using a syringeand 200-µm-diameter surgery needle. The syringe was removed prior to compression andmeasurement.

The time constant of the thermocouple was calibrated following the proceduredescribed in the paper of Park and Ro (1996). There may have been some errors in themeasurements owing to the thermal resistance between the droplet and the thermocouple

Dow

nloa

ded

by [

Kor

ea A

dvan

ced

Inst

itute

of

Scie

nce

& T

echn

olog

y (K

AIS

T)]

at 1

9:10

30

May

201

4

AUTO-IGNITION OF n-HEPTANE DROPLET 915

Figure 1 (a) Schematic diagram showing the experimental apparatus. (b) Positions of the sensors in the RCM,along with the clearance and adjustment wheel.

wire. However, it has been reported (Harada et al., 2011) that the maximum error in thetemperature of a 50-µm-diameter droplet in such an arrangement is <10%. The standardlimit of error of thermocouple was about 0.4%.

The pressure inside the reaction chamber was measured using pressure transducers(Sensys, Inc., PMS). The positions of the pressure sensors are shown in Figure 1a; thepressure could be measured over the range of 1–70 bar. To minimize the thermal shock, thepressure transducers were installed in ports in the cylinder wall with a cooling accessory.

Pressure and temperature data were recorded for a period of 1 s with a resolution of1 ms. These data were acquired using a data acquisition (DAQ) system (IOtech Inc., PDAQ3000 series) and stored using a PC.

Dow

nloa

ded

by [

Kor

ea A

dvan

ced

Inst

itute

of

Scie

nce

& T

echn

olog

y (K

AIS

T)]

at 1

9:10

30

May

201

4

916 H. KIM ET AL.

Optical Observation Setup

A high-speed CCD array camera was used to image the droplets. The ignition pointwas determined by selecting the image in which a yellow flame first appeared, and theignition delay was determined by counting the number of frames between a reference pointand the ignition point. A detailed discussion of the ignition point, ignition delay, and choiceof reference time is given in the next section.

The images were processed using a Visual Basic script to extract the temporal vari-ation in the diameter of the droplet. This software has been used in a previous study, andrepresents an established post-processing system (Ghassemi et al., 2005, 2006). The dropletboundary was determined from the location of the dark pixels in the image. The area of thedroplet was measured by counting the number of pixels inside the boundary. The diameterof the droplet was then calculated by introducing an effective diameter, whereby the areaof a circle with that diameter was equal to the area of the droplet in the image. A relativereference color was used as a parameter to control the detection of the droplet boundary.Further details of the post-processing system can be found in reference (Ghassemi et al.,2005).

It may be difficult to determine the interface between a droplet and vapor, whichcan appear blurred due to changes in the refractive index at the droplet/vapor interface.To mitigate these effects, we carried out preliminary experiments to determine the optimumcolors for determining droplet boundaries. A light-emitting diode (LED) backlight with afilter was used to avoid thermal radiation effects and oscillations in the brightness of thelight source.

Properties of the Fuel

n-Heptane is a standard reference, and is used to determine the octane number offuels. The boiling point of n-heptane droplets inside a reaction chamber is pressure depen-dent. To calculate the boiling point under all conditions found in the RCM, the KoreaThermophysical Properties Data Bank (KDB) was used. The boiling point can be foundover a range of pressures, and at temperatures over the range of −90.44◦C to +267.26◦C,by using the following relationship:

ln (pkPa) = A ∗ ln (Tboil) + B

Tboil+ C + D ∗ Tboil

2 (1)

where the coefficients of the equation are: A = −14.12388, B = −8030.070, C = 108.1461,and D = 1.204855 × 10−5.

The autoignition temperature is an important parameter in ignition delay. However,it is difficult to describe a standard value for the autoignition temperature, because it isaffected by the concentration of the fuel vapor, dimensions of the container, and pressure.Measured autoignition temperature values for n-heptane available in the literature vary, andlie within the range of 220–240◦C at normal pressure values (Drysdale, 1997). There isvery limited information on n-heptane autoignition temperature in the elevated pressure.Nevertheless, it is known that autoignition temperature usually decreases as the pressurerises, such that in the case of n-heptane, autoignition temperature dropped to 190◦C at10 bar (Brandes et al., 2005).

Dow

nloa

ded

by [

Kor

ea A

dvan

ced

Inst

itute

of

Scie

nce

& T

echn

olog

y (K

AIS

T)]

at 1

9:10

30

May

201

4

AUTO-IGNITION OF n-HEPTANE DROPLET 917

RESULTS AND DISCUSSION

Experimental Conditions

The initial droplet diameter and the operation time of the RCM were selected basedon references (Kim et al., 2011; Tsue et al., 2006). It is not straightforward to observeconditions inside a reaction chamber; however, to analyze droplet ignition phenomena ina meaningful manner it is essential to resolve the time-dependent pressure and tempera-ture. Thus, a numerical approach was used to examine the conditions inside the reactionchamber. Numerical simulations were carried out to analyze the variation in temperatureinside the reaction chamber. The initial diameter of the droplet was varied over the rangeof 400–1000 µm to observe its effects on droplet ignition. The ignition of the droplet wasevaluated for two different compression rates at the same final compression ratio.

The experiments were carried out at room temperature (i.e., 18◦C ± 2◦C) andatmospheric initial pressure. Before each experiment, the reaction chamber was chargedwith dry air to remove any fuel vapor and combustion products remaining from previousexperiments.

Determination of Ignition

Precise determination of the ignition point is important. Previous reports have usedthe appearance of a flame (Khan et al., 2007) or a change in the gradient of the temper-ature profile (Tanabe et al., 1995) to determine the ignition point. In RCM experiments,the ignition point is typically determined from changes in the pressure (Vranckx et al.,2013). However, it was not possible to measure pressure changes inside the reaction cham-ber because of the small quantity of fuel used in the experiments. Therefore, the ignitionpoint was determined optically by observing the appearance of a flame.

When a droplet is exposed to a high-temperature environment, some delay occurswhile the droplet is heated, evaporates, and it vapor mixes with the surrounding air; this istermed the physical ignition delay. Following this physical delay, ignition of a cool flameoccurs with a relatively low-temperature chemical reaction, and there is a further delaybefore a hot yellow flame appears. This is termed the chemical delay. There are severalreports of such a two-stage combustion process for hydrocarbons in relatively low ambienttemperatures (Khan et al., 2007; Tanabe et al., 1995). The total ignition delay is definedby summation of the physical delay and the chemical delay. In this study, the chemicaldelay was neglected, because it was not possible to observe the cool flame. Furthermore,the chemical delay is considerably shorter than the physical delay in a high-pressure envi-ronment (Stauch et al., 2006; Tanabe et al., 1996). The total ignition delay was thereforemeasured, and was defined as the time between the start of compression and the appearanceof a flame.

During the experiments, most of the droplets became detached from the thermocoupleduring combustion due to the decreased surface tension in the high-temperature environ-ment. Droplet temperature data following droplet detachment were therefore removed fromthe results.

Conditions in the RCM

The time variation of the temperature was calculated numerically. The compressionrate was controlled by varying the driving pressure. With a driving pressure of 12 bar and

Dow

nloa

ded

by [

Kor

ea A

dvan

ced

Inst

itute

of

Scie

nce

& T

echn

olog

y (K

AIS

T)]

at 1

9:10

30

May

201

4

918 H. KIM ET AL.

a compression ratio of 18:1, the compression stroke time was 235 ± 3 ms, and the peakpressure inside the reaction chamber was 28.7 bar. At a driving pressure of 18 bar, thecompression stroke time was 185 ± 2 ms and the peak pressure was 29.8 bar.

ANSYS Fluent v.13 was used to simulate the conditions inside the reaction chamber.The motion of the piston was first deduced following the procedure described in Mittaland Sung (2006) and Mittal et al. (2008). Figure 2 shows the pressure evolution insidethe reaction chamber, together with the position of the piston. The motion of the pis-ton was approximated using piecewise-polynomial fitting. The reaction chamber gridwas two-dimensional and axisymmetric, with 69000 nodes. The temperature-dependentthermophysical properties, including the specific heat capacity and viscosity, were takenfrom the reference (Poling et al., 2000). The pressure implicit split operator (PISO) algo-rithm and the k-ε equation for turbulent fluid flow were used (Mittal and Sung, 2006).Fixed temperature condition (291◦C) was assigned to wall boundary to simulate the heatloss through the cylinder wall.

To assess the validity of the numerical data, a comparison between the calculatedand measured pressure histories was carried out. Figure 3 shows the time variation of theexperimentally measured and simulated pressures. The overall trends of the results werein good agreement; however, there was some discrepancy near TDC, which is consistentwith reference (Mittal et al., 2008). One of the reasons for this is the truncation error thatoccurs when the piston motion is approximated by a polynomial. The pressure near TDCis very sensitive to the motion of the piston, so the approximation of the piston motioncan be a source of error. Moreover, the geometry of the sensor port and the quartz windowwere not considered in the 2-D axisymmetric model. The maximum error between theexperimentally measured and simulated pressures was 2.5%.

Figure 4 shows the variation of the maximum air temperature with time and thevolumetric averaged temperature inside the reaction chamber. The temperature increasedrapidly during compression. The difference between the maximum and volumetric averaged

0 100 200 3000

50

100

150

200

Pre

ssu

re (

bar)

Po

sit

ion

of

pis

ton

(m

m)

Time (ms)

Operation time

185 ms

235 ms

0

10

20

30

Figure 2 Time dependence of the pressure and piston position.

Dow

nloa

ded

by [

Kor

ea A

dvan

ced

Inst

itute

of

Scie

nce

& T

echn

olog

y (K

AIS

T)]

at 1

9:10

30

May

201

4

AUTO-IGNITION OF n-HEPTANE DROPLET 919

0 100 200 300 400

0

10

20

30

Pre

ss

ure

(b

ar)

Time (ms)

Experimental result

185 ms

235 ms

Numerical result

185 ms

235 ms

Figure 3 Time variation of the simulated and measured pressures.

0 100 200 300 4000

100

200

300

400

Tem

pera

ture

(°C

)

Time (ms)

Operation time : 185 ms

Maximum

Average

Operation time : 235 ms

Maximum

Average

Figure 4 Time variation of the maximum and volume-averaged temperatures inside the reaction chamber.

temperatures also increased during compression, which implies that heat loss occurred dur-ing compression. The maximum temperature was maintained at the center of the reactionchamber, where the droplet was located; the temperature in the radial direction decreasedcloser to the cylinder wall. For an operation time of 185 ms, the maximum temperature was383◦C, whereas that for an operation time of 235 ms was 351◦C. It follows that more heatloss occurred during the 235-ms compression stroke. The mean temperature was approx-imately 50◦C lower than the maximum temperature due to heat loss through the wallsof the combustion chamber. After compression, the maximum and volumetric averagedtemperatures were almost the same for both compression rates.

Dow

nloa

ded

by [

Kor

ea A

dvan

ced

Inst

itute

of

Scie

nce

& T

echn

olog

y (K

AIS

T)]

at 1

9:10

30

May

201

4

920 H. KIM ET AL.

Droplet Behavior

Figure 5 shows the time evolution of the droplet at a compression ratio of 18:1 anda compression time of 185 ms. A small increase in droplet diameter was observed becauseof thermal expansion (compare the images in Figures 5a and 5b). The droplet did not ignitefor a further 60 ms after TDC because sufficient fuel vapor had not yet been generated forignition to occur. Evaporation occurred due to the increased temperature inside the reactionchamber caused by the rapid pressure change. At 244 ms after the start of compression, aflame was observed near the droplet surface, which is characteristic of droplet ignition ina high-pressure environment. In high-pressure environments, mass diffusion processes aresuppressed, and so fuel vapor forms near the droplet (Khan et al., 2007). Following ignition,the flame propagates through the droplet. During droplet combustion, a luminous yellowflame covered the whole of the droplet due to the formation of soot. Approximately 40 msfrom the onset of ignition, the droplets typically became detached from the thermocouplewire owing to the decreased surface tension.

Figure 6 shows the time variation of droplet temperature, air temperature, and theboiling point of the fuel, which was calculated using Eq. (1), together with the dropletdiameter. The experiment for pure evaporation was done in nitrogen environment. Thedroplet temperature data for combustion applied only to cases where the droplet did notdetach from the thermocouple. In the early stages of compression, the air temperature roserapidly. It exceeded the auto-ignition temperature and boiling point of n-heptane, and con-tinued to rise until TDC. Consequently, the droplet temperature also increased; however,the rate was not as fast as that of the air temperature because of the differences in theheat capacity. The droplet temperature was lower than the boiling point during compres-sion period since the boiling temperature of n-heptane increased as the pressure inside thechamber rose. The droplet diameter was observed to increase during compression due tothermal expansion. Following compression, the air temperature decreased due to the heatloss through the cylinder walls. The droplet temperature however continued to rise, evenafter TDC was reached, because the droplet temperature was less than the boiling point and

a) b) c)

d) e)

Figure 5 Images of the droplet: (a) at the start of compression (0 ms), (b) after 184 ms (i.e., at the end ofthe compression stroke), (c) after 244 ms (at the time of ignition), (d) after 258 ms (corresponding to flamedevelopment), and (e) after 290 ms (detachment of the droplet).

Dow

nloa

ded

by [

Kor

ea A

dvan

ced

Inst

itute

of

Scie

nce

& T

echn

olog

y (K

AIS

T)]

at 1

9:10

30

May

201

4

AUTO-IGNITION OF n-HEPTANE DROPLET 921

0 100 200 300 4000.4

0.5

0.6

0.7Droplet temperature

Air temperature Pure evaporation

Boiling temperature Combustion

Dro

ple

t d

iam

ete

r (m

m)

Te

mp

era

ture

(°C

)

Time (ms)

Droplet diameter

0

200

400

End of combustion

Ignition point

End of compression

Figure 6 Time variation of the air temperature, boiling point, droplet temperature, and droplet diameter.

also less than that of the surrounding air. Based on these results, it can also be estimatedthat the temperature distribution inside the droplet may not be uniform, i.e., the surfacetemperature of droplet may be always higher than that of inner droplet until the occurrenceof autoignition. In addition, time scales for droplet heating and propagation of temperatureare considered to be quite longer than that of compression stroke, which is a distinctivecharacteristic compared to the droplet in steady condition.

When the droplet temperature reached 110◦C at about 230 ms, a rapid rise in thetemperature of the droplet was observed owing to the ignition. From this, it can be deducedthat a sufficient concentration of fuel vapor for ignition was generated during that period.The droplet temperature reached boiling point at approximately 290 ms after the start ofcompression. It is interesting to note that the droplet temperature gradually decreased afterit had reached boiling point because the boiling point decreased with the pressure insidethe reaction chamber following TDC.

It is also useful to compare the droplet temperature profile for combustion with thatof a pure evaporation process. Nitrogen gas was used to suppress ignition of the droplet.The time dependence of the droplet temperatures for ignition (with air) and pure evapora-tion (with nitrogen) are shown in Figure 6. These data reveal that there were no significantdifferences in the temperature profiles prior to ignition. In both cases, the temperature con-tinuously increased during the compression stroke. A rapid change in temperature wasobserved due to ignition of the droplet when air was present. When nitrogen was used, thetemperature of the droplet increased continuously until it reached about 145◦C at 450 ms.In the figure it can be found that the droplet is ignited during the heating process before thedroplet reaches its boiling temperature.

Effect of Droplet Initial Diameter

Figure 7 shows ignition images for initial droplet diameters of 553 µm, 723 µm,and 846 µm. The ignition delay time was 214 ms, 248 ms, and 266 ms, respectively. Thisclearly shows that there were differences in the ignition delay that were dependent on the

Dow

nloa

ded

by [

Kor

ea A

dvan

ced

Inst

itute

of

Scie

nce

& T

echn

olog

y (K

AIS

T)]

at 1

9:10

30

May

201

4

922 H. KIM ET AL.

a) b) c)

Figure 7 Images showing ignition for (a) a 553-µm-diameter droplet, (b) a 723-µm-diameter droplet, and (c) an846-µm-diameter droplet.

initial droplet diameter; as the droplet diameter increases, so does the ignition delay. Theseresults are consistent with previous reports for steady-state conditions (Tanabe et al., 1996).Regardless of the initial droplet diameter, a yellow flame is initiated at the lower edge of thedroplet, because in a gravitational environment n-heptane vapor is to migrate downwardssince its density is higher than that for air (Bergeron and Hallett, 1989).

In contrast to the droplet behavior at atmospheric pressure, the location of ignitionbecomes closer to the droplet in all cases (Stauch and Maas, 2007) in the high-pressure envi-ronment. Segawa et al. (2000) reported a decrease in the ignition delay at high pressuresdue to a decrease in the chemical reaction time (Stauch et al., 2006; Tanabe et al., 1996).Moreover, the mass diffusion processes of fuel vapor are slower in a high-pressure envi-ronment than at atmospheric pressure. Based on these considerations, it can be expectedthat ignition will commence as soon as fuel vapor is generated in the vicinity of thedroplet. Thus, the physical delay associated with vapor formation is the primary factorin determining the ignition delay in the experiments reported here.

The temporal variation of the droplet temperature is shown in Figure 8 for variousinitial droplet diameters. The droplet temperature started to rise at the start of compression.As the initial droplet diameter increased, the rate of the temperature rise decreased. Thereason for this is the increased heat capacity of the larger droplets, and the fact that thesurface area to volume ratio of a spherical droplet is inversely proportional to its diameter.Thus, the temperature of a smaller droplet rises faster. As the droplet temperature increases,a larger portion of the heat from the air is used for evaporation. Furthermore, the surfacetemperature for smaller droplet may be higher than that of larger droplet due to a highersurface to volume ratio, so that it would maintain higher fuel vapor concentration aroundthe droplet surface. Eventually, a sufficient concentration of fuel vapor is generated, whichwould result in ignition.

When the 402-µm-diameter droplet is ignited, the temperature of the droplet is114◦C. This shows that even if the droplet is not sufficiently heated, which is estimatedby the increasing trend of droplet temperature even at ignition, the fuel vapor concentrationaround the droplet becomes high enough for ignition. The droplet temperature at the igni-tion point is slightly lower for a larger diameter droplet, and it is 95◦C for an initial dropletdiameter of 843 µm. This is because for a larger droplet, sufficient fuel vapor may be gen-erated and accumulated at the exterior of the droplet even though heat transfer process intothe inner droplet is incomplete compared to smaller droplet.

Figure 8 also shows the temporal variation of the normalized droplet diameter, i.e.,the droplet diameter divided by the initial droplet diameter. These data show that smallerdroplets expanded more than larger ones because of differences in the droplet temperature.

Dow

nloa

ded

by [

Kor

ea A

dvan

ced

Inst

itute

of

Scie

nce

& T

echn

olog

y (K

AIS

T)]

at 1

9:10

30

May

201

4

AUTO-IGNITION OF n-HEPTANE DROPLET 923

Diameter of droplet

402 µm

578 µm

843 µm

945 µm

0 100 200 3000.95

1.00

1.05

1.10

Te

mp

era

ture

(°C

)

No

rma

lize

d d

rop

let

dia

me

ter

Time (ms)

End of compression

Ignition point

0

100

200

300

Figure 8 Time variation of the droplet temperature and normalized droplet diameter for different initial dropletdiameters.

In all droplets, ignition is observed to occur during thermal expansion. It follows thatsufficient fuel for ignition evaporated during heating of the droplets.

It is interesting to note that there is a critical limit to the droplet diameter, for whichignition did not occur. As shown in Figure 8, when droplet diameter reaches 945 µm,ignition does not occur. The term ‘maximum ignition limit’ is introduced to represent thisphenomenon. The reason for this is that sufficient vapor does not evolve during the exper-iment; instead most of heat is used for droplet heating. As discussed above, some portionof the heat flux is used to increase the droplet temperature, whereas the other portion isused for evaporation. For larger droplets, the surface area to volume ratio is too small toallow sufficient heat flux to be used for enough evaporation during the compression stroke.In other words, unless sufficient fuel vapor concentration is generated during the restrictedcompression time, the droplet would not be ignited. In an internal combustion engine, suchunburned fuel droplet would lead to low efficiency and pollutant gas emissions.

Figure 9 shows the ignition delay for various initial droplet diameters. These dataclearly show a monotonic relationship between the initial droplet diameter and the ignitiondelay, which is 188 ms for an initial droplet diameter of 400 µm and increases to 298 ms foran initial droplet diameter of 920 µm. Considering that the effect of mass diffusion processfrom liquid droplet to gas is dominant for droplet ignition, the main cause of this differenceis attributed to smaller value of surface to volume ratio for a larger droplet.

Effect of Compression Rate

The compression rate was controlled by varying the driving pressure. The compres-sion times were 235 ± 3 ms and 185 ± 2 ms, respectively. The temperature histories forthe two different compression rates for similar droplet diameters (578 µm and 593 µm)are shown in Figure 10. The rate of droplet heating was faster for the 185-ms compressiontime. Accordingly, the expansion rate of the droplet diameter was also higher for the 185-ms case. At the point of ignition, the droplet temperatures were almost the same. It follows

Dow

nloa

ded

by [

Kor

ea A

dvan

ced

Inst

itute

of

Scie

nce

& T

echn

olog

y (K

AIS

T)]

at 1

9:10

30

May

201

4

924 H. KIM ET AL.

400 600 800 1000150

200

250

300

350

Operation time : 185 ms

Compression ratio : 18Ig

nit

ion

dela

y (

ms)

Initial droplet diameter (µm)

Ignitable range

Maximumignition limit

End of compression

Figure 9 Ignition delay for different initial droplet diameters.

0 100 200 3000.95

1.00

1.05

1.10

Te

mp

era

ture

(°C

)

No

rma

lize

d d

rop

let

dia

me

ter

Time (ms)

Operation time

185 ms (578 µm)

235 ms (593 µm)

End of compression

Ignition point

185 ms 235 ms

0

100

200

300

Figure 10 Time variation of the droplet temperature and normalized droplet diameter for different compressionrates.

that certain levels of droplet heating are required for ignition to occur, regardless of the rateof gas phase heating by compression. For the 235-ms compression time, a slower rate oftemperature change in the gas and the droplet resulted in later ignition.

Figure 11 shows the ignition delay for the different compression rates. The ignitiondelay increased with the initial droplet diameter for both compression rates. The ignitiondelay for the 185-ms compression stroke was shorter than that of the 235-ms compressionstroke for all droplet diameters. It is due to the faster rise in air temperature and pressure asshown in Figure 4. As presented in Figure 10, its faster increasing rate of air temperature

Dow

nloa

ded

by [

Kor

ea A

dvan

ced

Inst

itute

of

Scie

nce

& T

echn

olog

y (K

AIS

T)]

at 1

9:10

30

May

201

4

AUTO-IGNITION OF n-HEPTANE DROPLET 925

400 600 800 1000150

200

250

300

350

Ignitable

range

Initial droplet diameter (µm)

Ign

itio

n d

ela

y (

ms)

Operation time

185 ms

235 ms

185 ms

235 ms

235 ms

185 ms

End of compression

Figure 11 Ignition delay and compression ignition delay for different compression.

makes the droplet heating rate faster and enhances fuel vapor formation around the dropletsurface. In addition, the peak air temperature inside the reaction chamber is higher for the185-ms case, which is one of the causes for shorter ignition delay (Ghassemi et al., 2006;Khan et al., 2007; Stauch et al., 2006).

The range of initial droplet diameter for which ignition can occur is also shown inFigure 11. The maximum ignition limit for the 235-ms compression stroke was 1000 µm,which was larger than that for the 185-ms compression stroke (945 µm). This was attributedto the air temperature profile in the reaction chamber (see Figure 4). Despite the higherpeak temperature for the 185-ms compression stroke, the temperature inside the reactionchamber after TDC was slightly higher for the 235-ms case, which reached its maximumtemperature later. Thus, with the 235-ms compression stroke, the droplet was exposed to ahigh-temperature environment for a longer period, leading to greater potential for evapora-tion and hence droplet ignition, so that a larger droplet could be ignited during the longercompression stroke.

These results clearly demonstrate that a temporal variation in temperature and pres-sure profiles plays an important role when considering the autoignition of fuel droplets.A steady-state description of the combustion is not sufficient to describe the autoignitionphenomena in the rapid compression machine.

CONCLUSIONS

The autoignition characteristics of a single n-heptane droplet have been experimen-tally investigated inside an RCM. The effects of the rapidly changing temperature andpressure on ignition were examined. While the thermo-fluid dynamic conditions inside thereaction chamber were considered using numerical simulations, the experimental measure-ments were discussed by changing initial droplet diameters and compression rates. Themain results of this study may be summarized as follows.

Dow

nloa

ded

by [

Kor

ea A

dvan

ced

Inst

itute

of

Scie

nce

& T

echn

olog

y (K

AIS

T)]

at 1

9:10

30

May

201

4

926 H. KIM ET AL.

1. The peak pressure for the 185-ms compression stroke was 29.8 bar, and the peaktemperature was 383◦C. For the 235-ms compression stroke, the peak pressure was28.7 bar, and the peak temperature was 351◦C. The higher peak pressure and tempera-ture were achieved for the 185-ms case because less heat was lost to the surroundingsduring the compression stroke. However, the temperature after TDC was higher for the235-ms case.

2. The droplet temperature increased during compression, and continued to increase afterTDC was reached because of the temperature difference between the droplet and the airinside the reaction chamber. A significant thermal expansion of droplets was observedbefore ignition. A rapid rise in droplet temperature was detected following ignition dueto the heat added from the chemical reaction. The droplet temperature remained closeto the boiling point until the entire droplet was burnt.

3. For a given compression rate, the ignition delay increased with the diameter of thedroplet due to smaller surface area to volume ratio of the droplet, which results inslower heating of the larger droplet. The longer ignition delay was observed for slowercompression stroke because of slower increasing rate of air temperature inside thechamber.

4. There was a maximum droplet size above which ignition did not occur, because theheating time was too short to produce sufficient fuel vapor for the droplet to be ignited.The maximum droplet size was larger for the longer 235-ms compression stroke casebecause of the longer exposure to the high-temperature environment.

FUNDING

This research was supported by a grant from the LNG Plant R&D Center funded bythe Ministry of Land, Infrastructure and Transport of the Korean government.

REFERENCES

Akiyama, H., Nishimura, H., Ibaraki, Y., and Iida, N. 1998. Study of diesel spray combustion andignition using high-pressure fuel injection and a micro-hole nozzle with a rapid compressionmachine: Improvement of combustion using low cetane number fuel. JSAE Rev., 19, 319.

Bergeron, C.A., and Hallett, W.L.H. 1989. Autoignition of single droplets of two-component liquidfuels. Combust. Sci. Technol., 65, 181.

Borghesi, G., Mastorakos, E., Devaud, C.B., and Bilger, R.W. 2011. Modeling evaporation effects inconditional moment closure for spray autoignition. Combust. Theor. Model., 15, 725.

Brandes, E., Hirsch, W., and Stolz, T. 2005. Autoignition temperatures for mixtures of flammableliquids with air at elevated pressures. No. 218. Presented at the European Combustion Meeting2005, Louvain-la-Neuve, Belgium, April 3–6.

Drysdale. D. 1997. An Introduction to Fire Dynamics, Wiley, New York.Ghassemi, H., Baek, S.W., and Khan, Q.S. 2005. Evaporation of binary droplets at elevated pressure

and temperature. AIAA Paper 2005-353. Presented at the 43rd Aerospace Sciences Meeting andExhibit, Reno, Nevada, January 10–13.

Ghassemi, H., Baek, S.W., and Khan, Q.S. 2006. Experimental study on binary droplet evaporationat elevated pressures and temperatures. Combust. Sci. Technol., 178, 1031.

Guibert, P., Keromnes, A., and Legros, G. 2010. An experimental investigation of the turbulenceeffect on the combustion propagation in a rapid compression machine. Flow Turbul. Combust.,84, 79.

Dow

nloa

ded

by [

Kor

ea A

dvan

ced

Inst

itute

of

Scie

nce

& T

echn

olog

y (K

AIS

T)]

at 1

9:10

30

May

201

4

AUTO-IGNITION OF n-HEPTANE DROPLET 927

Harada, T., Watanabe, H., Suzuki, Y., Kamata, H., Matsushita, Y., Aoki, H., and Miura, T. 2011.A numerical investigation of evaporation characteristics of a fuel droplet suspended from athermocouple. Int. J. Heat Mass Transfer, 54, 649.

Hong, N.S., Kobiri, S., Kamimoto, T., and Enomoto, Y. 1991. High temperature diesel combustion ina rapid compression-expansion machine. SAE Technical Paper Series, 1.

Honnery, D., Nguyen, D., and Soria, J. 2013. Microdroplet evaporation under increasing temperatureconditions: Experiments and modeling. Fuel, 105, 247.

Khan, Q.S., Baek, S.W., and Ghassemi, H. 2007. On the autoignition and combustion characteristicsof kerosene droplets at elevated pressure and temperature. Combust. Sci. Technol., 179, 2437.

Kim, H.M., Kim, H.Y., and Baek, S.W. 2011. Investigation of peak pressure variation due tocompression conditions using RCM. Trans. Korean Soc. Mech. Eng. B, 35 197.

Kukkadapu, G., Kumar, K., Sung, C.-J., Mehl, M., and Pitz, W.J. 2012. Experimental and surrogatemodeling study of gasoline ignition in a rapid compression machine. Combust. Flame, 159,3066.

Lee, D., and Hochgreb, S. 1998. Rapid compression machines: Heat transfer and suppression ofcorner vortex. Combust. Flame, 114, 531.

Ma, J., Lü, X., Ji, L., and Huang, Z. 2008. An experimental study of HCCI-DI combustion andemissions in a diesel engine with dual fuel. Int. J. Therm. Sci., 47, 1235.

Mittal, G., and Sung, C.-J. 2006. Aerodynamics inside a rapid compression machine. Combust.Flame, 145, 160.

Mittal, G., Raju, M.P., and Sung, C.-J. 2008. Computational fluid dynamics modeling of hydrogenignition in rapid compression machine. Combust. Flame, 155, 417.

Park, S.J., and Ro, S.T. 1996. A new method for measuring time constants of a thermocouple wire invarying flow states. Exp. Fluids., 21, 380.

Poling, B.E., Prausnitz, J.M., and O’Connel, J.P. 2000. The Properties of Gases and Liquids,McGraw-Hill, New York.

Roth, F.A., and Heidelberg, N. 2000. Dynamics of Droplets, Springer, New York.Segawa, D., Kadota, T., Nakainkyo, A., Hirota, S., and Enomoto, H. 2000. Effect of ambient pressure

on autoignition of fuel droplet in supercritical and microgravity environment. Proc. Combust.Inst., 28, 1063.

Stauch, R., and Maas, U. 2007. Different ignition modes of n-heptane droplets in air. Presented at theThird European Combustion Meeting (ECM), Chania, Crete, April 11–13.

Stauch, R., Lipp, S., and Maas, U. 2006. Detailed numerical simulations of the autoignition of singlen-heptane droplets in air. Combust. Flame, 145, 533.

Tanabe, M., Bolik, T., Eigenbrod, C., Rath, H.J., Sato, J., and Kono, M. 1996. Spontaneous ignitionof liquid droplets from a view of non-homogeneous mixture formation and transient chemicalreactions. Symp. (Int.) Combust., 26, 1637.

Tanabe, M., Kono, M., Sato, J., Koenig, J., Eigenbrod, C., Dinkelacker, F., and Rath Zarm, H.J. 1995.Two stage ignition of n-heptane isolated droplets. Combust. Sci. Technol., 108, 103.

Tsue, M., Ishimaru, R., Ukita, T., Nakaya, S., Kono, M., and Okai, K. 2006. Spontaneous ignition offuel droplets in lean fuel-air mixtures. J. Propul. Power, 22, 1339.

Vranckx, S., Lee, C., Chakravarty, H.K., and Fernandes, R.X. 2013. A rapid compression machinestudy of the low temperature combustion of cyclohexane at elevated pressures, Proc. Combust.Inst., 34, 377.

Wang, T.J., and Baek, S.W. 2007. Numerical modelling of unsteady spray behavior in moderatelyhigh pressure regime. Combust. Theor. Model., 11, 21.

Watanabe, H., Suzuki, Y., Harada, T., Matsushita, Y., Aoki, H., and Miura, T. 2010. An experimentalinvestigation of the breakup characteristics of secondary atomization of emulsified fuel droplet.Energy, 35, 806.

Yang, J.-R., and Wong, S.-C. 2003. On the suppression of negative temperature coefficient (NFC) inautoignition of n-heptane droplets. Combust. Flame, 132, 475.

Dow

nloa

ded

by [

Kor

ea A

dvan

ced

Inst

itute

of

Scie

nce

& T

echn

olog

y (K

AIS

T)]

at 1

9:10

30

May

201

4