ILASS-Americas 25th Annual Conference on Liquid Atomization and Spray Systems, Pittsburgh, PA, May 2013

Experimental Characterization 1-Butanol Electrospray Phenomenology

Michael Pennisi1∗ Dimitrios C. Kyritsis123

1Department of Mechanical Science and Engineering,University of Illinois Urbana-Champaign, 1206 W. Green St., Urbana IL, 61801

2Department of Mechanical Engineering,Khalifa University of Science and Technology, and Research Abu Dhabi, UAE

3International Institute for Carbon Neutral Energy Research (WPI-I2CNER),Kyushu University, 744 Moto-oka, Nishi-ku, Fukuoka 819-0395, Japan

AbstractThis study presents the visualization and characterization of electrosprays of 1-butanol that burn in air.Spray phenomenology was visualized and features such as shape, size, and stability of the spray flame weredetermined. Phase Doppler Anemometry (PDA) and Mie-scattering-based droplet size measurements wereperformed in both reacting and non-reacting electrosprays. In general, a strong peak in the distributionof droplet diameters centered between 40 and 50µm was observed in the non-reacting case, with a secondsmaller peak observed around 200µm. In the reactive case, a broader size distribution was observed, stillwith a peak between 40 and 50µm, but also with a larger percentage of droplet diameters between 20 to50µm. The Mie scattering results were compared with PDA measurements taken at the center of the sprayfor several voltage and flow rate conditions. In both cases, Reynolds and Weber numbers were calculated andwere related to observed spray and flame phenomenology. In addition, velocimetry measurements providedinsight to the velocity of the droplets inside the burning spray. Some droplets were observed not to vaporizecompletely and passed through the flame. Photographic evidence was acquired of droplet fission in theburning spray. Charge transferred by the spray was also measured and compared with the Rayleigh limitcriterion for droplet fission. An analysis of the evaporation time, based on droplet size and velocity, was usedin order to rationalize the results. Indeed, large droplets traveling with the measured velocities could passthrough the flame without completely evaporating. It was concluded that the electrosprays of 1-butanol couldsustain flames stabilized with the assistance of electrostatics. Spray flames with electrostatically assistedatomization and stabilization were shown to have characteristics substantially different from the spray flamesof non-charged fuels.

∗Corresponding Author: pennisi2@illinois.edu

1 Introduction

Electrospray is an injection system which causesliquid atomization using electrostatic forces, withoutthe use of a large pressure differential. Previouselectrospray combustion research has been primarilyconducted with traditional hydrocarbon fuels, suchas heptane. Gomez and his collaborators atYale University performed studies with heptanein a counter flow burner, injecting through anelectrospray and studying the effect on the flat flameproduced in their configuration [1]. Specificallymentioned in that paper was the inability for directelectrospray combustion, citing the breakdown ofthe electric field because of the high concentrationof chemi-ions in the flame. Additionally, largerdroplets were observed to have a higher averagevelocity than smaller ones, and, therefore, had lowerresidence time in the flame [1]. In a subsequentstudy, the same group sought to detect and captureevidence of droplet fission in the flames of acounter-flow burner. Their theory sought to explainthe rapid loss of larger sized droplets, given thatevaporation effects were insufficient to account forthis finding. In several images, convincing evidencewas presented of abnormally close droplets in areaswith otherwise sparse populations [2]. This wouldsuggest that one larger droplet broke apart intoseveral smaller droplets as a result of Rayleigh Limitinduced fission. Shrimpton his collaborators [3]have performed numerous studies on electrosprayand droplet characteristics, most recently evaluatingReynolds and Weber number effects of both laminarand turbulent regimes on spray atomization. Thisgroup introduced a modified Weber number totake into account the effective reduction of surfacetension in the presence of the electric field [3].

Electrospraying is particularly appropriate forbio-alcohols that have an electric conductivity ofat least four orders of magnitude higher than theones of hydrocarbons. With recent advances inlarge scale bio-butanol production [4, 5], butanolmay become a mainstream transportation fuel in thefuture. Butanol has many advantages over ethanol,which is currently the most popular liquid biofuel.In addition to having a higher energy density thanethanol (36.4 vs 24.8 MJ/kg), butanol requiresless molecular breakdown during production, whichtranslates into less energy needed in production.Furthermore, butanol can be easily transported inpipelines and it can be mixed with diesel fuel withoutthe need for additives. Promising combustionbenefits with the use of electrosprays have beendemonstrated in enhancing fuel injection in internalcombustion engines [6, 7], and power generation [8]

with a particular emphasis on micro-combustion [9,10]. Research into electrospray enhancement of fuelinjectors for use with ethanol-gasoline and butanol-gasoline blends has been conducted [7, 11], showingbenefits in emission reduction. More recentlybutanol electrosprays have been compared to thoseof ethanol and conductivity enhanced heptane byAgathou and Kyritsis [12, 13].

Published literature lacks fundamental studiesof butanol electrospray combustion. With thisstudy, we would like to offer some initial exper-imental data about this combustion phenomenon.Flame phenomenology was described, and the effectof the presence of the flame on defining parametersof the spray, such as droplet size and velocity,was examined and compared to previous studies onelectrospray flames. Of particular interest was thedual effect of electrostatics on both atomization andflame stabilization.

2 Experimental Apparatus

The experimental apparatus is shown schemat-ically in Figure 1. The electrospray is formed byconnecting a steel capillary connected to a syringepump and high voltage source. The capillary had aninner diameter of 127µm and a conical tip that wasmachined using an Electrical Discharge Machine toenhance the electric field at the nozzle.

The flow of liquid butanol ran opposite togravity, and it was metered with a KD-Scientificsyringe pump. n-Butanol was used in all of the testsbecause this is the dominant isomer in bio-butanolmixtures. The distance between the capillary tipand grounded mesh was fixed at 12.5 cm for alltests. Voltages of 6.5-7.5kV and flow rates in therange of 10-20 ml/hr were found to produce self-sustaining air-stabilized flames. An air-stabilizedflame is defined as a flame that burns at a substantialdistance above the capillary tip, anchored at alocation that is essentially determined by the fuelflow rate at a given voltage. At each flow rate, thepotential field was varied at 6.5, 7.0 and 7.5 kV.However, the important parameter is the potentialdivided by the distance between the cathode and theanode. With the fixed gap of 12.5 cm, this translatesto 520, 560, and 600 V/cm respectively. At eachvoltage setting, the butanol flow rate was varied at10, 15, and 20 ml/hr.

Droplet size measurements were obtained usinga Malvern Spraytec 97 droplet sizer. This deviceutilizes a 670 nm laser and detector system. Also,a Dantec PDA system with the capability of two-dimensional velocity measurements was used tocapture both droplet diameter and velocity. The

2

514.5 nm Laser

532 nm Laser

PDA Detector Steel Capillary

Cylindrical Lens and

resulting laser sheet

Butanol Electrospray

(Blue)

SLR/High Speed

Camera

Figure 1. Laser sheet through the air-stabilized flame

system was equipped with a BSA P60 processorconnected to 58N70 detector. The laser used in thePDA system was a Spectra Physics Stabilite 2017Argon-ion.

3 Electrospray Phenomenology

The factor that determined the mode in whichthe electrosprays combusted was the ratio betweendroplet spacing and droplet size. As the dropletsbecame bigger and spread farther apart, thesingle droplet combustion mode was preferred (Fig.2). Large droplet spacing inhibited the flamesof individual droplets from merging together andforming a larger flame fueled by droplets burningin the group combustion mode.

If the droplets are small and close to eachother, the droplets burn as a group (left imagein Fig. 3), with an intermediate state beinginternal group combustion (right image in Fig. 3).Clearly, electrostatic control can drastically affectflame morphology. Furthermore, in all of the testconditions, some of the droplets either deviated awayfrom the flame before reaching it or passed at adistance from the flame that prevented completeevaporation and subsequent ignition.

Figure 4 shows a non-burning spray at 600V/cm. Under this applied potential, an almostdual spray structure appears, that is much morepronounced than in lower voltage cases. Specifically,a narrow central core spray (red arrows) and a muchbroader outer spray (blue arrows) are clearly visible.When the flame burns in the mode shown on theleft panel in Fig. 3, it burns on the fuel from the

Figure 2. Single droplet combustion mode withlaser sheet present

inner spray cone, whereas the much broader flameon the right panel of Fig. 3 received fuel from boththe inner and outer spray cones. The existence oftwo cones is consistent with the findings of Tangand Gomez [14], who observed what they termed“satellite droplets” in their flames.

3

Figure 3. Different combustion modes for 20 ml/hrbutanol flow rates and a 7.5 kV applied potential

4 Droplet Size Analysis

Droplet size measurements were performed asclose to the base of the flame as possible withouthaving the measurement volume inside of the flame,so that beam-steering effects were avoided. Thiswas 18 mm from the tip of the capillary. Manycombinations of electric field and flow rate weretested; however, only results acquired with a fieldintensity 560 V/cm and mass flow rate of 15ml/hr case are presented for the purpose of brevity.Probability density functions of droplet size areshown in Fig. 5 on the following page with andwithout a flame present. The probability densityof a particular droplet volume is presented asa function of the diameter corresponding to thisvolume. Significant differences can be seen betweenthe two cases. The size distribution when theflame is not present is notably different from themonodisperse sprays observed in previous studies inthe cone-jet mode [2, 1].

In the reacting case, the distribution becomesmuch more peaked around a central value - in thiscase about 42µm. Both the number of dropletslarger and smaller than this central size havedecreased, most likely a result of evaporation ofthe droplets as they approach the high temperatureregion of the flame. This would reduce the size ofthe largest droplets and nearly eliminate the smallestdroplets.

The velocity component along the spray axisversus droplet diameter is presented in Fig. 6.

Inner Cone

Outer Cone

Figure 4. Electrospray Structure: 20 ml/hrbutanol flow rate and 7.5 kV applied potential, noflame present

A wide range of velocities for each droplet size isobserved; however, larger droplets on average havea lower velocity. During combustion, the dropletswith the median size were measured to have a largerrange of velocities than during the non-combustingcase. This could be due to the large dropletsevaporating but remaining at the same velocity asthey approach the high temperature region of theflame. Furthermore, these droplets have the largestsurface area and would vaporize at a faster rate thansmaller droplets, as described in equation 4.

Droplet diameters were also measured usingthe Mie-scattering-based Malvern Spraytec. TheMalvern takes a line average across the test region,unlike the essentially point measurement of thePDA. The Malvern also captured data at 18 mmfrom the capillary tip. It is further noted that,the Malvern results are volume-based PDFs, so adirect comparison between the output of the twoinstruments should be made carefully.

In the non-combusting e-sprays, the majority ofthe spray volume is contained in droplets between20 and 60µm using both measurement devices.Interesting differences between the two measurementdevices do exist. Both droplets that are smaller andlarger than the smallest and largest sizes measuredin the PDA experiments are present. This couldbe due to droplets that have escaped the electricfield and fall back under gravity could be measuredagain using the Malvern, increasing their volumepercentage. This is not a issue encountered with thePDA because the measurement location was chosento be at the center on the spray.

Furthermore, when measuring near the e-spray

4

Figure 5. Probability density function withoutcombustion (top) and with combustion (bottom)

flame, the droplet size distribution broadens acrossdroplets sizes less than 100µm when using theMalvern but becomes narrower using the PDA.Previous work by Yule et al. [15] has suggested thatthe droplets on the periphery of electrosprays areusually smaller and highly charged. Photographicdata indicated that the droplet spacing in acombusting e-spray was lower than in the non-combusting e-sprays. In these more compactsprays, the larger droplets exist near the spayaxis, whereas smaller ones are near the periphery.Since the measurement location of the PDA isfixed, it recorded a much higher percentage oflarger-sized droplets. However, the line-integratedmeasurement of the Malvern continued to captureall of the droplets sizes present. Furthermore,a reduction in droplet size should be expectedas the droplets approach the high temperatureregion near the flame. Because of the reductionof droplet spacing, the PDA should only recorda reduction in the maximum size of the droplets,because the largest droplets are located at the coreof the spray. Comparing both the non-combustingand combusting e-spray data taken with the PDA,the largest droplet size is reduced from 65µm toabout 50µm during combustion. Data capturedwith the Malvern reveals a reduction of all dropletsizes and a broadening of the entire distribution,possibility due to the different rates of evaporationof droplets at different locations in the spray and

with varying initial diameters. Another explanationof the reduction of the droplet diameters could bedroplet fission, which will be discussed in detail insection 5.

In the Malvern data, droplets diameters around300µm were measured with and without a flame.More detailed probing of both combusting and non-combusting electrosprays with the PDA is underwayat this time to help provide a more conclusive theoryas to the differences between the measurementsystems In particular, the size and velocity of thedroplets on the periphery of the e-spray.

The captured flame images and droplet velocitydistributions reveal two important characteristics ofthese burning electrosprays. First, droplets are notall completely vaporized by the time they reach theflame. Second, droplets moving at high speed (>10m/s) are observed inside the flame envelope that

Figure 6. Comparison between Droplet Diameterand Vertical Velocity (non-combusting top, com-busting bottom)

5

Figure 7. Log-scale PDFs of droplet volumemeasured with the Malvern instrument: Non-combusting e-spray (top) and combusting e-spray(bottom)

surrounds the spray. Following the analysis of Turns[16], we provide a plausible explanation for dropletspassing through the flame. In particular, estimatingthe thermal conductivity of the gaseous butanol andair mixture kg as

kg = 0.4 ∗ kC4H10O+ 0.6 ∗ kAir (1)

Also, Bq is the ratio of the sensible enthalpy in thevaporized butanol to the enthalpy of vaporization

Bq =Cpg ∗ (T∞ − Tboil)

Hfg(2)

T∞ is the adiabatic flame temperature for a butanol-air flame, Tboil is the boiling temperature for butanolat 1 atm, Cpg is the specific heat of the butanolvapor, and Hfg is heat of vaporization of butanol.

The evaporation constant “K” can be thenestimated as:

K =8kg

ρC4H10O∗ Cpg

∗ ln(Bq + 1) (3)

In such a case, the evaporation time tD is given by,where D is the initial droplet diameter:

tD =D2

0

K(4)

The results of this calculation are presented inTable 1, based on an observed flame size of 6.5cm. The critical speed, as defined here, is themaximum speed at which a droplet can travelingto ensure complete vaporization before leaving theflame region. Therefore, droplets traveling fasterthan this velocity would pass through the flamebefore vaporizing completely.

DropletDiameter(µm)

EvaporationTime (ms)

CriticalSpeed(m/s)

50 4.76 13.755 5.76 11.360 6.86 9.4865 8.05 8.0870 9.33 6.9675 10.1 5.33

Table 1. Evaporation time and critical speed tocompletely vaporize

Indeed, 65-75µm droplets moving faster thanthe calculated critical speed were observed at thebase of the spray flames. Thus, this simplemodel provided a plausible justification of theobservations of droplets passing through the flamebefore completely vaporizing.

5 Droplet Fission

Strong evidence of droplet fission inside andabove the electrospray flames was acquired throughhigh-speed visualization, shown in Fig. 8. Thisparticular image was taken at 520 V/cm electricfield with 20 mL/hr butanol flow rate. Gomez etal. [2] theorized droplet fission was a mechanismof reducing droplet size, but did not providephotographic evidence of the phenomenon. In ourvisualization data, it was occasionally observed thatone droplet would be replaced by two or threedroplets slightly downstream of the first in theimmediately following frame. The parent dropletwould first enlarge before separating into multiplesmaller droplets. It is conceivable that dropletswould have some out of plane velocity, i.e. movingin and out of the laser sheet, and would producethis same effect. However, several instances offission occurred with few other droplets around.Furthermore, the newly formed droplets were veryclose together, within two pixels of each other(yellow box in Fig. 8 on the next page), thissupporting the assertion that multiple droplets wereproduced from one larger droplet and that thesewere not droplets entering the imaging plane. Figure

6

8 provided strong photographic indication of dropletfission.

Figure 8. Apparent droplet fission in a combustinge-spray

In order to check our hypothesis, we compare thecharge carried by the droplet with the theoreticallypredicted Rayleigh limit for droplet fission [17].

q2 = 8π ∗ ε0 ∗ γB ∗D3 (5)

Where q is the surface charge density, ε0 = 8.8542 ∗10−12 C2m−2N−1 [18] is the permittivity of freespace, γB is the liquid surface tension of butanol(23.96 ∗ 10−3 N/m) [19], and D is the diameter ofthe droplet. As the droplet evaporates, the chargedensity on the surface of the droplet increases untilthe Rayleigh limit is surpassed and the dropletbreaks apart into smaller droplets. The followingtable lists the Rayleigh limit for typical dropletdiameters found in the butanol electrosprays of thisstudy

To compute the critical droplet diameterspecified by the Rayleigh criterion, measurementsof the charge transferred to the spray were alsotaken. For the 520 V/cm electric field and 20ml/hr flow rate, the measured charge transfer was0.12 nanoamps. With this information, and giventhe known butanol volume flow rate of 20 mL/hr,the charge injected to the liquid butanol was foundto be 21.6 C/m3. For this charge density, thecritical droplet diameter of approximately 51µm(using equation 5. Droplets larger than this areunstable according to the Rayleigh criterion, andthis would suggest that steep decrease in the numberof droplets between 50 and 60µm is a result of fissionbecause the droplet charge exceeded Rayleigh limit.

DropletDiameter(µm)

RayleighLimit (C)

(C/m3)

10 7.301E-14 139.520 2.065E-13 49.3130 3.794E-13 26.8440 5.841E-13 17.4360 1.073E-12 9.48970 1.352E-12 7.52980 1.652E-12 6.16390 1.971E-12 5.165100 2.309E-12 4.410

Table 2. Rayleigh limit of butanol for severaldroplet diameters

6 Summary and Conclusions

Reactive bio-butanol electrosprays can operateboth in the the individual droplet combustionmode and group-combustion. Changing the appliedelectric field and fuel flow rate resulted in changingthe stable position of the spray flame as well asthe droplet combustion mode of the spray. Inthe non-reacting electrospray, an almost bi-modaldroplet size distribution was observed, but duringcombustion, the numbers of both the largest andsmallest sized droplets were reduced.

Furthermore, the comparison of droplet velocitywith and with out a flame present revealed a narrow-ing of the droplet velocity with a flame present. Thedifferences in the droplet size PDF’s between a pointsource and line-of-sight measurement devices werecompared on the basis of decreasing droplet spacingin the combusting e-spray. Combining the measureddroplet diameter and velocity information, a simpledroplet evaporation model was used to suggestthat some droplets could pass through the flameregion without completely vaporizing. Furthermore,images of apparent droplet fission in a combustinge-spray were justified using a Rayleigh limit basedanalysis.

7 Acknowledgments

The authors would like to acknowledge thesupport of the National Science Foundation throughgrant CBET-12-36786, Dr. Ruey-Hung Chen,Program manager. DCK gratefully acknowledgesthe support of the International Institute forCarbon Neutral Energy Research (WPI-I2CNER),sponsored by the World Premier InternationalResearch Center Initiative (WPI), MEXT, Japan.

7

References

[1] G. Chen and A. Gomez. Proceedings of theCombustion Institute, 24(1):1531 – 1539, 1992.

[2] A. Gomez and G. Chen. Combustion Scienceand Technology, 96(1-3):47–59, 1994.

[3] A. Kourmatzis, E. L. Ergene, J. S. Shrimpton,D. C. Kyritsis, F. Mashayek, and M. Huo.Experiments in Fluids, 53(1):221–235, JUL2012.

[4] T. Eseji, N. Qureshi, and H.P. Blaschek.Process Chemistry, 42:34–39, 2007.

[5] Y. Tashiro, H. Shinto, H. Hayashi, S. Baba,G. Kobayashi, and K. Sonomoto. Biosci.Bioeng, 104:238–240, 2007.

[6] E.K. Anderson, A. P. Carlucci, A. de Risi,and D.C. Kyritsis. Atmomization and Sprays,17:289–313, 2007.

[7] E.K. Anderson and D. C. Kyritsis. 20th AnnualInstitute for Liquid Atomization and SpraySystem, Chicago, IL, 2007. ILASS-AmericasConference.

[8] M.S. Agathou, J.W. Powell, C.F. Lee, and D.C.Kyritsis. SAE Paper No., 24-0020:–, 2007.

[9] D. C. Kyritsis, B. Coroton, F. Faure, S. Roy-choudhury, and A. Gomez. Combustion andFlame, 139:77–89, 2004.

[10] D. C. Kyritsis, S. Roychoudhury, C. Mc Enally,C. Pfefferle, and A. Gomez. Experimental Fluidand Thermal Science, 28:97–104, 2004.

[11] E. K. Anderson and D. C. Kyritsis. The5th Joint Meeting of the U.S. Sections of theCombustion Institute, San Diego, CA, 2007.Paper G11.

[12] Maria S. Agathou and Dimitrios C. Kyritsis.Energy Conversion and Management, 2012.

[13] Maria S. Agathou and Dimitrios C. Kyritsis.Energy Conversion and Management, 2012.

[14] K. Q. Tang and A. Gomez. Journal of Colloidaand Interface Science, 184:500–511, 1996.

[15] A.J. Yule, J.S. Shrimpton, A.P. Watkins,W Balachandran, and D Hu. Fuel, 74(7):1094–1103, JUL 1995.

[16] Stephen R Turns. An Introduction to Combus-tion: Concepts and Applications, pp. 378–395.WCB/McGraw-Hill, 2000.

[17] Lord Rayleigh. Philosophical Magazine Series5, 14(87):184–186, 1882.

[18] National Insittute of Standards and Technol-ogy. Nist reference on constants, units, anduncertanity, June 2011.

[19] Anna Wandschneider, Jochen K. Lehmann,and Andreas Heintz. Journal of Chemical &Engineering Data, 53(2):596–599, 2008.

8