optically accessible high-pressure combustion apparatuscoesdytse/papers/2004-rsi_apparatus.pdfof a...

Optically accessible high-pressure combustion apparatusStephen D. Tsea)

Department of Mechanical and Aerospace Engineering, Rutgers, the State University of New Jersey,Piscataway, New Jersey 08854

Delin Zhu and Chung K. LawDepartment of Mechanical and Aerospace Engineering, Princeton University, Princeton, New Jersey 08544

~Received 24 July 2003; accepted 27 October 2003!

The design and operation of a novel optically accessible high-pressure combustion apparatus ispresented. The apparatus provides optical access for the direct observation of the morphology anddevelopment of premixed reaction fronts at elevated pressures. A chamber-in-chamber design withan innovative connecting system allows for safe, constant-pressure measurements, alleviating theextreme overpressures encountered in high-pressure combustion processes within closed bombs.Auxiliary design features include gap-adjustable electrodes for spark ignition and ports for jetstirring. As a result, the apparatus is well suited for the study of laminar premixed flames, flameinstabilities, turbulent flames, and detonations. Results from the study of centrally ignited hydrogenand methane fuels in oxygen-inert mixtures up to 60 atm initial pressure demonstrate the suitabilityof the apparatus for high-pressure combustion experiments. ©2004 American Institute of Physics.@DOI: 10.1063/1.1634358#

I. INTRODUCTION

Recognizing that combustion processes within internalcombustion engines take place in elevated pressure environ-ments, that most fundamental information on the aerother-mochemistry of flames have been obtained at substantiallylower pressures, and that the chemical reaction mechanismswhich govern the various flame processes are inherently non-linear, which can cause nonmonotonic variations of the com-bustion responses with pressure, it is necessary to obtain fun-damental understanding and useful data on flame propa-gation, extinction, and stability at pressures that are tens ofatmospheres, and hence are representative of those in inter-nal combustion engines. Of particular interest is the determi-nation of the laminar flame speeds,su

0, of fuel–air mixtures,which are needed for the modeling of processes within inter-nal combustion engines and also for the development ofcomprehensive oxidation reaction mechanisms of these fuelsthat are valid at such high pressures.1

The counterflow flame technique2 has recently been ex-tensively used in the determination ofsu

0 of a large number offuels under a variety of conditions including elevated pres-sures. The maximum pressure attainable, however, has beenlimited to 6 or 7 atm. The difficulty is that the flow becomesunstable as its Reynolds number increases with increasingpressure. Although this effect can be somewhat mitigated byusing nozzles of small diameters, there is a lower limit in thenozzle size because it has to be of a similar magnitude as thenozzle separation distance, which in turn is restricted by the

flame thickness and radial dimension so that the flame canstill be considered to be planar.

An alternate means of measuringsu0 is by using the

spark-ignited, outwardly propagating flame in an enclosedchamber. In the original approachsu

0 is indirectly determinedthrough the temporal increase of the chamber pressure,3 as-suming that the flame is smooth and spherically symmetric,and the propagation is steady. However, without direct ob-servation of the flame configuration, the validity of theseassumptions cannot be scrutinized. Furthermore, flamestretch effects4 also cannot be accounted for.

Groff5 was the first to measure spherical flame speeds athigh pressures~up to 5 bar! by directly imaging the propa-gating flame front. The propensity of cell formation over theoutwardly propagating flame front was reported. Smith andco-workers6–8 and Faeth and co-workers9–14 have deter-mined su

0 for various fuel compositions and pressuresthrough systematic subtraction of stretch effects using simi-lar imaging techniques. The values so determined, with dif-ferent stretch–compensation relations, are in good agreementwith each other, as well as with the counterflow experiments.However, these experiments have been limited to pressureslower than about 5 atm; and laminar flames, including thoseaffected by various instabilities, have not been examined insuch a sufficiently rigorous manner for conventional gaseousfuels at elevated pressures up to the range of 50 atm.

In view of the above considerations, we have designedand developed an experimental apparatus suitable for thestudy of high-pressure spherical flame propagation with op-tical access up to 60 atm. Because of the usefulness andpotential versatility of a design of this nature for the study ofhigh-pressure combustion phenomena in general, we reportherein the design, test procedure, and representative results.

a!Author to whom correspondence should be addressed; electronic mail:[email protected]

REVIEW OF SCIENTIFIC INSTRUMENTS VOLUME 75, NUMBER 1 JANUARY 2004

2330034-6748/2004/75(1)/233/7/$22.00 © 2004 American Institute of Physics

Downloaded 20 Jan 2004 to 128.6.73.23. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/rsio/rsicr.jsp

http://dx.doi.org/10.1063/1.1634358

II. EXPERIMENTAL DESIGN

A. Dual-chamber apparatus

Figure 1 shows the schematic of the apparatus. It con-sists of two concentric aluminum~alloy 6061, temper T6511!cylindrical vessels. The inner vessel~82.55 mm ID, 107.95mm OD, 127 mm length! is initially filled with the combus-tible premixture under testing, and the outer chamber~273.05 mm ID, 323.85 mm OD, and 304.8 mm length! isinitially filled with an inert gas, e.g., nitrogen/helium, at thesame pressure. For optical access, a pair of quartz windows~101.6 mmB, 6.35 mm thick! comprises the two circularends of the inner vessel, and a pair of quartz windows~114.3mm B, 25.4 mm thick! is mounted within the two endflanges~393.7 mmB, 38.1 mm thick! of the outer chamber.Eight carbon–steel threaded rods~19.05 mmB! connect thealuminum end flanges, holding the pressure within. The ap-paratus has been hydrostatically tested to an operating pres-sure of 95 atm. A 1300 psi check valve ensures safety.

The lateral walls of the inner vessel and an encasingcylindrical sleeve~107.95 mm ID, 114.3 mm OD, 127 mmlength! contain matching rows of holes, which when offset,seal via O rings the inner vessel from the surrounding outerchamber~Fig. 1!. In synchronization with ignition, the en-casing sleeve of the inner vessel is mechanically translated,in a piston-like manner, such that the rows of holes betweenthe inner vessel and its encasing sleeve align, establishingcontinuity between the inner and outer vessels. There areeight circumferential rows of 54 holes~3.6 mmB! totaling432 holes, affording minimal pressure drop between the twovessels. Since the volume ratio of the outer to inner vessel is

about 25:1, the total pressure increase in the entire systemdue to combustion within the inner vessel is small, measuredto be less than 3%, hence, allowing near-constant-pressureexperimentation and ensuring operational safety.

The plumbing diagram of the apparatus, involving gasdelivery, pressure measurement, mixing, venting, and evacu-ation, is shown in Fig. 2. Further details are provided in thefollowing sections. One important design aspect of the appa-ratus is the use of nonrotating stem valves for all pipingconnections leading to the inner vessel filled with the com-bustible. It was noticed that the ball valves, in the process ofbeing opened/closed, could ignite the premixture especiallyat high pressures, presumably due to frictional heating be-tween the metal ball element and the Teflon valve seat.

During the filling process, a digital pressure gauge~Ce-Comp DPG1000DR! monitors the pressure in either the in-ner or outer vessel, and a differential pressure transducer~SenSym 19U050G4! monitors the pressure difference be-tween the two vessels. The magnitude of the pressure differ-ence between the inner and outer vessels must be limited to10 psi in order to prevent breaking of the quartz windowsenclosing the inner vessel. Pressure relief check valves con-nect the two vessels in both directions to prevent over/underpressuring the inner windows, as shown in Fig. 2. Byusing one pressure gauge and the differential pressure trans-ducer, the pressure in both chambers can be ‘‘equalized’’during filling.

After filling, the reactants within the inner vessel arethoroughly mixed by jet stirring through a dual-port injectionand suction technique powered by a two-way, piston-driven

FIG. 1. Schematic of the dual-chamber apparatus. The two inner chambers in the figure represent when the holes on the inner vessel and encasing sleeve areoff set ~left!, and when the holes are aligned~right!. As can be seen, the mechanical lever, which translates the encasing sleeve, also triggers the spark andhigh-speed camera.

234 Rev. Sci. Instrum., Vol. 75, No. 1, January 2004 Tse, Zhu, and Law


pump~Fig. 2!. Turbulent mixing takes place inside the innerchamber, where suction and discharge jets~;30 mL/s! alter-nate between the inlet ports~4 mm ID!. Sufficient mixing isqualitatively assessed by schlieren imaging, as the densitygradients in the jet reduce as the components become bettermixed, and quantitatively assessed by gas chromatograph~GC! sampling. AK-type thermocouple measures the tem-perature of the initial combustible mixture prior to ignition.

B. Ignition

Ignition energy is deposited by a spark at the center ofthe inner vessel via electrodes extending from opposite loca-tions on the lateral surface. One electrode is fixed while theother can be moved with a micrometer, providing an adjust-able gap to within 10mm. The tips of the electrodes are finetungsten wires, having a diameter of 250mm and a freelength of 25 mm. Spark energy is supplied by a combinationof a high-voltage capacitive discharge into an ignition coil~20 kV!, to break down the gap, and a constant high-voltagedc power supply~3 kV!, to sustain the spark, producing ap-proximately square voltage and current profiles. The durationof the spark is adjustable~5 ms–1 ms! via a timing circuit intandem with a floating-ground power metal–oxide–semiconductor field-effect transistor, controlling the deliveryof current from the high-voltage power supply. Voltage andcurrent from the ignition pulse are recorded onto a 200 MHzdigital oscilloscope to evaluate the amount of energy depos-

ited. Spark energies were adjusted to be close to the mini-mum ignition energies~0.5–20 mJ!, in order to minimizeeffects of initial flame acceleration by excessive spark ener-gies. The gap distance is usually set from 0.3 to 1.0 mm,depending on the working pressure. Note that the plumbingfor the electrodes can be replaced with small windows toallow for laser ignition.

C. Optical diagnostics

Measurements involve observing the flames usingschlieren/shadowgraph motion picture photography. Forcompactness~for eventual microgravity experimentation!, anaxially transmissive schlieren/shadowgraph configuration isutilized. In the system, a 100 W mercury short-arc lamp isfocused by a 50 mm camera lens to an image of 0.5 mm,which is spatially filtered by a 100mm pinhole. A 250 mmachromatic lens~76.2 mm B! collimates the light, whichpasses through the apparatus through the quartz windows. A250 mm achromatic lens~76.2 mmB! then focuses the col-limated light onto a horizontal, adjustable knife edge. With a50 mm camera lens, the schlieren image is recorded by ahigh-speed digital camera~Phantom V7.0! from 1000 to100 000 fps with exposure times from 1 to 10ms, adjusted tomaximize image brightness and sharpness. Shadowgraph im-ages are taken by removing the knife edge and defocusingthe camera lens in front of the high-speed camera so that theimage no longer corresponds to the confocal plane. The

FIG. 2. Plumbing diagram of the high-pressure combustion apparatus.

235Rev. Sci. Instrum., Vol. 75, No. 1, January 2004 High-pressure combustion apparatus


schlieren system is also tuned for a given pressure to com-pensate for the slight lensing effect from the outer quartzwindows due to pressure-induced curvature.

D. Pressure history

A 0–1000 psia pressure transducer with a 0.1 ms re-sponse time~SenSym 13U1000AM! measures the absolutepressure within the chamber during the combustion event. A200 MHz digital oscilloscope records the pressure history.

III. OPERATIONAL PROCEDURE

A. Filling

Filling of the chamber is done with the inner and outervessels sealed from each other such that there is no continu-ity of the gases. Note that the two chambers are filled con-currently in order to prevent rupture of the inner vessel orbreaking of the inner windows, as mentioned in the forego-ing discussions. Additionally, the vessels are filled veryslowly to minimize entropic heating of the gases, which af-fect the pressure reading.

The reactant mixtures are prepared within the inner ves-sel by adding individual component gases~e.g., hydrogen,oxygen, nitrogen, and helium! using the partial pressuremethod. The molar fraction of each component is calculatedfor a given equivalence ratio, and its partial pressure can beestablished for the given total pressure of the experiment.Since oxidizer is used to vent the chamber between experi-ments, as will be explained later, the oxidizer is the firstcomponent to be introduced into the inner chamber~but onlypartially!. Near atmospheric levels, the component with thehighest compressibility factor is introduced in order to mini-mize composition errors due to real gas compressibility ef-fects at high pressures. The final mixture is brought up to thefinal pressure with the least compressible component. Theuse of certified oxidizer mixtures~oxygen1inert! assures re-peatability in the mixture composition, hence, removing onepossible source of error during the filling process. A sampleoperational procedure is given in Table I~see Fig. 2 for ref-erence!.

Note that a purging procedure is needed whenever thefilling process switches from oxidizer to fuel and vice versa.If one needs to switch the gas feed to the inner chamber,NV-1 is closed first when the desired pressure is reached, andthen 3BV-1 switches from fuel to oxidizer, and vice versa.The volume trapped between 3BV-1 and NV-1, if not purged,will be introduced into the inner chamber, altering the mix-ture composition. This is especially significant for a fuel likemethane, due to the stoichiometric ratio with the oxidizer.Therefore, the procedure of repeatedly using the vacuum lineto purge the trapped gas and filling the volume with the new

FIG. 3. Effect of density differences between the inner and outer chambers:~a! equal densities,~b! small magnitude of difference in densities, and~c!large magnitude of difference in densities.

TABLE I. Valve positions during an experimental cycle.

Fillingstep

Description andcomments Closed valvesa Open valvesa

1 Chambers readyfor filling

NRV-1,8 NRV-2,3,6,7

Initial settings~two chambersisolated!

NV-1,23BV-1,2,3,4

NRV-4 ~or 5!b

BV-1c

2 Initial oxidizer 3BV-1 to oxidizerfilling 3BV-4 to N2 /He

NRV-1, NV-1

3 Oxidizer to fuelchange

NRV-1, NV-1 3BV-2 to vacuum3BV-1

4 Fuel filling 3BV-2 to fuel NRV-1,NV-1

5 Fuel to oxidizerchange

NRV-1, NV-1 3BV-2 to vacuum

6 Final oxidizer filling 3BV-2 3BV-1 to oxidizerNRV-1, NV-1

7 Mixing period~approximately10 min!

NRV-1, NV-13BV-1,4

8 Decay period~approximately10 min!

3BV-4 and NV-1,or NV-2, to equalizepressure

9 Ready for ignition NRV-2,3,4,5,6,7 NRV-8NV-1,2, BV-13BV-4

10 Pressure release NRV-2,3,5,6,7~two chambersconnected!

BV-1c

3BV-3 to vent

11 First vacuum NRV-5 NRV-43BV-3 to vacuum

12 Oxidizer vent~fill the chamberto 1.0 psia!

3BV-3 3BV-1 to oxidizerNRV-1

13 Second vacuum NRV-13BV-1

3BV-3 to vacuum

14 End 3BV-3

aClosed and open valves remain so in further steps unless otherwise indi-cated.

bClose NRV-5 and open 4 when subatmospheric, close NRV-4 and open 5when superatmospheric.

cBV-1 should be closed before the outer chamber pressure reaches the gaugemaximum.



gas ensures that only the selected gas will be delivered to theinner chamber when NV-1 is opened again.

Simultaneously, the outer vessel is filled with inert gases,which will later absorb the pressure release from the inner

vessel during combustion. An inert mixture is prepared in theouter vessel with nitrogen~high molecular weight! and he-lium ~low molecular weight! in calculated proportions suchthat the average density of the mixture in the outer vessel is

FIG. 4. Outward flame propagation displaying:~a! smooth, spherical flames for flame speed extraction;~b! diffusional-thermal instability;~c! hydrodynamicinstability; and~d! hydrodynamic wrinkling cascading down to small scales. Note:a5O2 /(O21N2).



the same as that in the inner vessel. This action is critical inreducing buoyant flow between the two vessels once conti-nuity between them is established. For example, if the den-sity of the mixture in the inner vessel is much less than thatin the outer vessel, upon connecting them by aligning theholes, the combustible mixture will rush upward out of theinner vessel, being replaced by inert rushing into it frombelow from the outer vessel. As such the experiment can beso severely compromised that no combustion is initiated.Figure 3 shows the effect of nonmatching densities.

B. Mixing

After the chambers are charged, the mixing system isactivated~Fig. 2!. Experiments show that approximately 10min duration is necessary to produce homogenous mixtures.Again, homogeneity can be qualitatively inspected via theschlieren images. After mixing, the mixture is allowed tosettle in order to enable all fluid motion to dissipate vis-cously, as well as for the temperature to equilibrate to that ofthe surroundings. The final mixture within the inner chamberis systematically sampled by GC and analyzed for certifica-tion.

C. Combustion

To initiate combustion, the inner vessel’s encasing sleeveis then translated through a lever action, establishing conti-nuity between the vessels~Fig. 1!. The same lever actiontriggers the high-speed camera as well as the spark, and thequiescent combustible mixture is ignited. Flame propagationis terminated when the flame reaches the walls of the innervessel, with the total pressure buildup attenuated due to thelarge outer-vessel volume. As such, near-constant-pressurecombustion takes place.

D. Postexperiment

After the experiment, the chamber is vented of the high-pressure gas, and a vacuum pump removes all combustionproducts and inert gases from the apparatus. To ensure mini-mum contamination, the vessels are then partially filled withthe oxidizer mixture up to 1.0 psia, and then repurged byvacuum. As such, any residual gas would be largely oxidizer,which is the first component to be introduced into the innervessel for the next experiment.

IV. RESULTS AND DISCUSSION

Figure 4~a! shows a characteristically smooth, spheri-cally symmetric, outwardly propagating flame from whichflame speed data can be extracted. As a benchmark for thereliability of the apparatus and its experimental results,stretch-corrected laminar burning velocitiessu

0 as a functionof equivalence ratio for hydrogen–air mixtures at standardtemperature and pressure~STP! were extracted and com-pared with numerical simulation, and with the stretch-corrected results from Taylor and co-workers6,7 and Faethand co-workers.12 The extracted flame speeds are given inRef. 15, agree within experimental uncertainties, and are re-peatable.

At high pressures, there is a propensity of cell formationover the flame surface due to hydrodynamic and thermal-diffusive instabilities, even despite the stabilizing effects ofpositive flame stretch. This is because at high pressures, theflame thickness decreases, promoting both the thermal-diffusive and hydrodynamic instabilities. Figure 4~b! showsan outwardly propagating (Le,1) flame at 10 atm display-ing the diffusional-thermal instability, characterized bythermal-diffusive cells of the dimension of the flame thick-ness. Figure 4~c! shows an outwardly propagating,diffusional-thermally stable (Le.1) flame at 10 atm display-ing the hydrodynamic instability, acting at all wavelengths.Such flame ‘‘cracks’’ are expected to appear initially in largedimensions, as shown in Fig. 4~c!, being triggered bysystem-based perturbations, and cascade down to smallercells until they reach a size similar to that of the thermal-diffusive cells when they are stabilized by the curvature, asseen in Fig. 4~d! at 60 atm.

Nonetheless, through manipulation of the transport prop-erties, inherent instabilities can be either mostly suppressedor delayed. As a result, stretch-free laminar flame speedshave been determined up to 20 atm for hydrogen15 and 60atm for methane,16 and are compared with the calculatedvalues allowing for detailed chemistry and transport.

The device is also well suited for examining criticalflame phenomena~e.g., extinction, flame balls, cool flames!at high pressures. However, buoyancy distorts the flameshape, especially for weakly propagating, near-limit flames,as seen in Fig. 5. Since buoyant forces increase with increas-ing pressure, microgravity facilities become requisite tomaintain spherical symmetry when the characteristic time forflame propagation is of the order of that for buoyant convec-tion.

FIG. 5. Buoyancy effect on lean methane/air flame at 3 atm under 1 g.



The optically accessible high-pressure combustion appa-ratus allows one to measure the burning and flammabilityproperties of combustible mixtures of single and mul-tiphases. The novel design allows for the safe testing of com-bustible mixtures at high pressures, at which data for mostpractical fuels are needed but have not been tested. Thesedata are acquired at the scientific/engineering level of rigor,permitting one to obtain fundamental understanding of flamepropagation, extinction, flammability, and stability, and tocompile comprehensive reaction mechanisms for practicalfuels presently utilized in power generation and propulsionsystems at high pressures. In addition, our technique goesmuch farther beyond the measurement of laminar flamespeeds; it offers the means for in-depth study of other inter-esting combustion phenomena, such as ignition, flame insta-bility, self-turbulization, and even deflagration-to-detonationtransition, under high pressures.

ACKNOWLEDGMENT

This work was supported by NASA under its Micrograv-ity Combustion Program.

1C. K. Law, C. J. Sung, H. Wang, and T. F. Lu, AIAA J.41, 1629~2003!.2C. K. Wu and C. K. Law, Proc. Combust. Inst.20, 1941~1985!.3M. Metghalchi and J. C. Keck, Combust. Flame38, 143 ~1980!.4C. K. Law and C. J. Sung, Prog. Energy Combust. Sci.26, 459 ~2000!.5E. G. Groff, Combust. Flame48, 51 ~1982!.6D. R. Dowdy, D. B. Smith, S. C. Taylor, and A. Williams, Proc. Combust.Inst. 23, 325 ~1990!.

7S. C. Taylor, Ph.D. thesis, University of Leeds~1991!.8M. J. Brown, I. C. McLean, D. B. Smith, and S. C. Taylor, Proc. Combust.Inst. 26, 875 ~1996!.

9S. Kwon, L.-K. Tseng, and G. M. Faeth, Combust. Flame90, 230 ~1992!.10L.-K. Tseng, M. A. Ismail, and G. M. Faeth, Combust. Flame95, 410

~1993!.11M. I. Hassan, K. T. Aung, and G. M. Faeth, J. Propul. Power13, 239

~1997!.12K. T. Aung, M. I. Hassan, and G. M. Faeth, Combust. Flame109, 1

~1997!.13K. T. Aung, M. I. Hassan, and G. M. Faeth, Combust. Flame112, 1

~1998!.14M. I. Hassan, K. T. Aung, and G. M. Faeth, Combust. Flame115, 539

~1998!.15S. D. Tse, D. L. Zhu, and C. K. Law, Proc. Combust. Inst.28, 1793

~2000!.16G. Rozenchan, D. L. Zhu, C. K. Law, and S. D. Tse, Proc. Combust. Inst.

29, 1461~2002!.



optically accessible high-pressure combustion apparatuscoesdytse/papers/2004-rsi_apparatus.pdfof a...

Documents