Experimental and numerical investigation of the hetero...

Available online at www.sciencedirect.comProceedings

Proceedings of the Combustion Institute 32 (2009) 1947–1955

www.elsevier.com/locate/proci

of the

CombustionInstitute

Experimental and numerical investigationof the hetero-/homogeneous combustion

of lean propane/air mixtures over platinum

Symeon Karagiannidis a, John Mantzaras a,*, Rolf Bombach a,Sabine Schenker a, Konstantinos Boulouchos b

a Paul Scherrer Institute, Combustion Research, CH-5232 Villigen-PSI, Switzerlandb Swiss Federal Institute of Technology, Laboratory of Aerothermochemistry and Combustion Systems,

CH-8092 Zurich, Switzerland

Abstract

The pure heterogeneous and the coupled hetero-/homogeneous combustion of fuel-lean propane/airmixtures over platinum have been investigated at pressures 1 bar 6 p 6 7 bar, fuel-to-air equivalence ratios0.23 6 u 6 0.43, and catalytic wall temperatures 723 K 6 Tw 6 1286 K. Experiments were performed in anoptically accessible catalytic channel-flow reactor and involved 1-D Raman measurements of major gas-phase species concentrations across the reactor boundary layer for the assessment of catalytic fuel conver-sion and planar laser induced fluorescence (LIF) of the OH radical for the determination of homogeneousignition. Numerical predictions were carried out with a 2-D elliptic CFD code that included a one-step cat-alytic reaction for the total oxidation of propane on Pt, an elementary C3 gas-phase chemical reactionmechanism, and detailed transport. A global catalytic reaction step valid over the entire pressure–temper-ature-equivalence ratio parameter range has been established, which revealed a �p0.75 dependence of thecatalytic reactivity on pressure. The aforementioned global catalytic step was further coupled to a detailedgas-phase reaction mechanism in order to simulate homogeneous ignition characteristics in the channel-flow reactor. The predictions reproduced within 10% the measured homogeneous ignition distances at pres-sures p 6 5 bar, while at p = 7 bar the simulations overpredicted the measurements by 19%. The overallmodel performance suggests that the employed hetero-/homogeneous chemical reaction schemes are suit-able for the design of propane-fueled catalytic microreactors.� 2009 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

Keywords: Catalytic combustion of propane on platinum; Hetero-/homogeneous combustion; Pressure dependence ofcatalytic reactivity; Homogeneous ignition

1540-7489/$ - see front matter � 2009 The Combustion Institdoi:10.1016/j.proci.2008.06.063

* Corresponding author. Fax: +41 56 3102199.E-mail address: [email protected] (J.

Mantzaras).

1. Introduction

Catalytic combustion is being pursued in largepower generation systems as an ultra-low-NOx

technology with enhanced flame stability undereither fuel-lean or fuel-rich operating conditions[1–3]. Methane, the main constituent of natural

ute. Published by Elsevier Inc. All rights reserved.

mailto:[email protected]

Fig. 1. Test rig and Raman/LIF setup.

1948 S. Karagiannidis et al. / Proceedings of the Combustion Institute 32 (2009) 1947–1955

gas, has been the fuel of interest in the aforemen-tioned studies. The catalytic combustion of higherhydrocarbons and of hydrogen has attractedincreased attention during the last years for theportable production of energy and heat usingmicro- and mesoscale power devices [4–6].

Propane is a fuel of particular interest formicro-energy conversion systems since it liquefiesat room temperature and moderate pressuresand is commercially available in compact contain-ers for numerous consumer applications. Thestudy of its complete oxidation over catalytic sur-faces constitutes a natural first step towardsunderstanding the similar behavior of higherhydrocarbons. Some of the physical characteris-tics of propane are common among higher hydro-carbons, such as the larger-than-unity Lewisnumber (a property directly impacting the catalystsurface temperatures [7,8]) and the negative tem-perature coefficient of the gas-phase ignition char-acteristics under certain operating conditions.

On a fundamental level, the oxidation of pro-pane over noble metals at atmospheric pressurehas been studied along with higher and lowerhydrocarbons [9,10], with the latter study employ-ing a non-activated dissociative chemisorptionstep for propane and a temperature-dependentadsorption/desorption reaction set for oxygen. Ithas been shown that the catalytic reactivity of lin-ear alkanes increases with increasing carbon chainlength, due to weaker CAC bond strengths ofhigher hydrocarbons; in addition, platinum isthe most active noble metal for the oxidation ofall alkanes except methane [11]. Detailed hetero-geneous reaction mechanisms for fuel-lean com-bustion of higher hydrocarbons have not yetprogressed to the same extent as those of methane[12,13]. Recent studies of fuel-lean propane oxida-tion over Pt-based catalysts suggested an overallreaction that is first order with respect to propaneand zero-th order with respect to oxygen [14].Accompanying experiments of propane catalyticcombustion have mainly been conducted at atmo-spheric pressure [4,6,15].

We have recently [16] investigated numericallythe stability maps of methane-fueled catalytic mic-roreactors at pressures up to 5 bar, a range ofinterest to microturbine-based power generationsystems. Detailed hetero-/homogeneous reactionschemes were used, validated over the appropriatepressure and temperature ranges with in situ laser-based spectroscopic measurements [17–19]. Theformer study [16] exemplified the importance ofgas-phase chemistry and of high operating pres-sure for extending the combustion stability limitsof microreactors, while the latter [17,18] have alsoaddressed the implications of using global cata-lytic reaction steps in modeling methane hetero-/homogeneous combustion under high pressures.

The present work undertakes a combinedexperimental and numerical investigation of the

catalytic combustion of fuel-lean C3H8/air mix-tures over platinum at pressures 1 6 p 6 7 bar, arange of interest to microreactors and small-scaleindustrial turbines. Experiments have been carriedout in an optically accessible, channel-flow cata-lytic reactor at propane-to-air equivalence ratiosranging between 0.23 and 0.43. The heterogeneousreactivity is assessed with Raman spectroscopy ofmajor gas-phase species concentrations across thechannel boundary layer, and the onset of homoge-neous ignition is monitored with planar laserinduced fluorescence (LIF) of the OH radical.Simulations are performed with a full elliptic 2-D CFD code. The main objectives are to assessthe catalytic reactivity of propane and its pressuredependence by constructing an appropriate globalreaction step and then to couple this step with adetailed homogeneous reaction mechanism so asto reproduce the measured homogeneous ignitioncharacteristics.

2. Experimental

2.1. High pressure test rig

The test-rig (Fig. 1) consisted of a channel-flowcatalytic reactor, which was mounted inside ahigh-pressure cylindrical tank. The reactor com-prised two horizontal Si[SiC] plates with a length(x) of 300 mm, width (z) of 110 mm and thicknessof 9 mm; the plates were positioned 7 mm apart(y). The other two sides of the reactor wereformed by two 3-mm-thick vertical quartz win-dows. The inner surfaces of the Si[SiC] plates werecoated via plasma vapor deposition with a 1.5 lmthick Al2O3 non-porous layer, followed by a2.2 lm thick platinum layer. BET and CO-chemi-sorption measurements verified the absence ofporous structure in the catalyst layer [17]. The sur-

S. Karagiannidis et al. / Proceedings of the Combustion Institute 32 (2009) 1947–1955 1949

face temperatures along the x–y symmetry planewere monitored with S-type thermocouples (12for each plate), which were embedded 0.9 mmbeneath the catalyst layer. To facilitate the cata-lytic reactivity studies, a kinetically controlled fuelconversion away from the mass-transport-limitwas necessary. To avoid transport limitationsoriginating from the high catalytic reactivity ofpropane over platinum, a coupled cooling/heatingarrangement was adopted to control the surfacetemperatures: in a fashion similar to earlier hydro-gen catalytic combustion studies [20], the reactorentry section was water-cooled to maintain alow catalytic reactivity while the rear of the reac-tor (x > 100 mm) was heated by two resistive heat-ers placed above the ceramic plates.

A compressor provided dry air, which was pre-heated and mixed with propane (Grade 3.5 purity)in two sequential static mixers. The preheatedpropane/air mixture was driven into the reactorthrough a 50-mm long inert rectangular honey-comb section that provided a uniform inlet veloc-ity profile. The reactor inlet temperature wasmonitored with a thermocouple placed down-stream of the honeycomb section. The high-pres-sure vessel was fitted with two 350-mm long and35-mm thick quartz windows (see Fig. 1), whichmaintained optical accessibility from both reactorsides. Two additional quartz windows located atthe exhaust section of the vessel and the reactoroutlet provided a counterflow optical access forthe LIF experiments. Apart from propane/air,experiments with propane/air/oxygen mixtureshave also been carried out.

2.2. Laser diagnostics

The LIF/Raman set-up is also shown inFig. 1. The particularly low volumetric contentof fuel in lean propane/air combustion necessi-tated the use of a dedicated high power laserfor the Raman measurements. The 526.5 nmradiation of a frequency-doubled Nd:YLF highrepetition rate pulsed laser (Quantronix DarwinDuo), operated at 1.5 to 2 kHz, with a pulseduration and energy of 130 ns and 37 to 43 mJ,respectively, provided the light source for theRaman measurements. Given the laminar andsteady operating conditions, the signal of20,000 to 40,000 pulses was integrated on thedetector chip when acquiring an image. Six ofthese images were subsequently averaged, suchthat up to 9 kJ of laser light was used for a singleRaman spectrum. The signal-to-noise ratio wasthus increased by a factor of 20 compared tothe previous methane/air Raman arrangement[18,19]; moreover, the danger of dielectric gasbreakdown was eliminated due to the resultinglower intensities at the focal line.

The 526.5 nm beam was focused through thetank and reactor side-windows into a vertical line

(�0.3 mm thick) by an f = 150 mm cylindricallens. The focal line spanned the 7 mm channel sep-aration and was offset laterally (z = 15 mm) toincrease the collection angle and minimize thermalbeam steering [19]. Two f = 300 mm lenses col-lected the scattered light at a 50� angle withrespect to the incident optical path and focusedit to the entrance slit of a 25 cm imaging spectro-graph (Chromex-250i) equipped with an intensi-fied CCD camera (Princeton Instruments PI-MAX1024GIII). The 1024- and 256-pixel-longCCD dimensions corresponded to wavelengthand transverse distance, respectively. The effectiveRaman cross sections, which included transmis-sion efficiencies, were evaluated by recording thesignals of pure propane, air, and completely burntgases of known composition. Raman data for themajor combustion species C3H8, H2O, N2, O2,and CO2 were acquired at different positions bytraversing axially a table supporting the sendingand collecting optics and also the Nd:YLF laser(Fig. 1). The 250-pixel-long 7 mm channel heightwas binned to 63 pixels.

For the OH-LIF, the 532 nm radiation of afrequency-doubled Nd:YAG laser (QuantelTDL90 NBP2UVT3) pumped a tunable dyelaser (Quantel TDL90); its frequency-doubledradiation (285 nm) had a pulse energy of 0.5mJ, low enough to avoid saturation of theA(v = 1) X(v0 = 0) transition. The 285 nmbeam was transformed into a laser sheet by acylindrical lens telescope and a 1 mm slit mask,which propagated counterflow, along the x–ysymmetry plane (Fig. 1). The fluorescence ofboth (1–1) and (0–0) transitions at 308 and314 nm, respectively, was collected at 90�(through the reactor and tank side-windows)with an intensified CCD camera (LaVision Ima-ger Compact HiRes IRO, 1392 � 1024 pixelsbinned to 696 � 512). A 120 � 7 mm2 sectionof the combustor was imaged on a 600 � 34pixel CCD-area. The camera was traversed axi-ally to map the 300 mm reactor extent; at eachmeasuring location 400 images were averaged.

3. Numerical

The lack of detailed surface reaction mecha-nisms for propane on platinum necessitated theuse of a single-step catalytic reaction coupled toa detailed gas-phase reaction mechanism. The glo-bal reaction step of Garetto et al. [14] has beendeveloped for the total oxidation of propane toH2O and CO2 over Pt at atmospheric pressure,for a range of equivalence ratios encompassingthe conditions of this work:

_sC3H8¼ A� T 1:15

w � expð�Ea=RT wÞ � ½C3H8�aw:ð1Þ


The catalytic rate _sC3H8is in [mol/cm2s],

A = 93.2 K�1.15 cm1.45 mol�0.15 s�1, Ea = 71.128 kJ/mol, a = 1.15 and the concentration of propane[C3H8] is in [mol/cm3]. Finally, the subscript wdenotes conditions at the gas–wall interface.

An optimized mechanism for homogeneouscombustion of C1–C3 species by Qin et al (70 spe-cies, 14 irreversible and 449 reversible reactions)[21] was employed for modeling gas-phase chemis-try. Thermodynamic data were included in theprovided scheme. Surface and gas-phase reactionrates were evaluated with Surface-CHEMKIN[22] and CHEMKIN [23], respectively. Mixture-average diffusion was the transport model, usingthe CHEMKIN transport database [24].

A steady, full-elliptic, 2-D laminar CFD codewas used in the simulations (for details see [19]).An orthogonal staggered grid of 450 � 140 points(x and y, respectively) was sufficient to give a gridindependent solution for the 300 � 7 mm2 channeldomain. Uniform inlet profiles were applied fortemperature, axial velocity and species mass frac-tions. The interfacial energy boundary conditionswere prescribed wall temperature profiles; theseprofiles were polynomial curves fitted throughthe thermocouple measurements of the upperand lower walls. No-slip was applied for bothvelocity components at the walls (y = 0 and7 mm), while zero-Neumann conditions were usedat the outlet.

4. Results and discussion

The experimental conditions are provided inTable 1. The inlet Reynolds numbers, ReIN,(based on the uniform inlet properties and thechannel hydraulic diameter) were as high as2171, leading to laminar flows; recent turbulentcatalytic combustion studies [25] have shown thatthe strong flow laminarization induced by the hot

Table 1Experimental conditionsa

Case P u TIN UIN ReIN N2 vol.%

1 1 0.30 398 1.45 827 552 3 0.27 400 1.10 1831 603 5 0.27 399 0.66 1836 604 7 0.26 398 0.55 2162 605 3 0.31 446 1.58 2171 776 5 0.23 448 0.94 2162 557 7 0.31 446 0.68 2170 778 1 0.43 492 1.39 544 779 3 0.42 468 0.51 655 77

10 5 0.43 478 0.28 571 7711 7 0.42 475 0.29 859 77

a Pressure, equivalence ratio, inlet temperature, veloc-ity, Reynolds number and nitrogen dilution. Cases with77% vol. N2 pertain to C3H8/air mixtures.

catalytic walls guarantees laminar flow conditionsat considerably higher ReIN. Cases 1–7, character-ized by the absence of homogeneous ignition, pro-vided the platform for evaluating the catalyticreactivity and its pressure dependence with theaid of the Raman measurements. In Cases 8–11,whereby flames were established inside the chan-nel, the gas-phase combustion processes wereevaluated using the OH-LIF data.

4.1. Effect of pressure on the catalytic reactivity

Comparisons between Raman-measured andpredicted transverse profiles of C3H8 and H2Omole fractions, at four selected streamwise loca-tions, are illustrated in Fig. 2 for Cases 1–4; forclarity, 27 of the total 63 measured points are pro-vided. Modest profile asymmetries are evident dueto small differences between the upper- and lower-wall temperatures (see the wall temperatures ofFig. 3). The experimentally resolved transverseextent ranged from 0.3 6 y 6 6.7 mm at high pres-sures down to 0.6 6 y 6 6.4 mm for the atmo-spheric pressure experiments (due to theincreasing Raman signal-to-noise ratio with risingpressure). The measurement accuracy was ±10%for compositions as low as 0.5% vol.; concentra-tions less than 0.5% vol. entailed larger measure-ment uncertainties. To further increase theRaman signal of the deficient reactant (propane),C3H8/air/O2 mixtures were also examined in cer-tain cases, whereby the volumetric content ofnitrogen was as low as 55%. Axial profiles of thecomputed catalytic (C) and gas-phase (G) propaneconversion rates (the latter integrated over the 7-mm channel height), are presented in Fig. 3 forCases 1–4. The maximum reactor extent overwhich the G conversion is negligible has beendelineated with the arrows in Fig. 3 (defined asthe positions where the G conversion amounts to5% of the C conversion). For the assessment ofthe catalytic reactivity, the forthcoming compari-sons of Raman data with predictions are limitedto this extent so as to avoid falsification of the cat-alytic kinetics by gas-phase chemistry. The initialreactor extent over which the gas-phase contribu-tion could be safely neglected ranged from10.9 cm (Case 3) to 14.2 cm (Case 4).

The heterogeneous reactivity of propane wasassessed by varying the reactor pressure and sur-face temperature, and monitoring the near-wallbending of the propane transverse profiles. InCase 2 (Fig. 2, p = 3 bar) the catalytic fuel conver-sion is already appreciable at surface temperaturesTw 6 850 K, while mass-transport-limited condi-tions are approached for Tw = 1064 K (mani-fested by the low propane mole fractions nearboth walls in Fig. 2(2d)). In contrast, previousCH4/air studies [18] in the same reactor necessi-tated an increase in pressure to achieve a similarlevel of reactivity under the same mass inflowcon-

Fig. 2. Measured and predicted transverse profiles of C3H8 and H2O mole fractions for Cases 1–4 at four streamwiselocations. Measurements: C3H8 (circles), H2O (triangles). Predictions: C3H8 (solid lines, pressure-corrected model;dashed lines, Garetto), H2O (dashed-dotted lines, pressure-corrected model; dotted lines, Garetto).


ditions. Numerical predictions using the globalcatalytic reaction rate of Garetto yielded a verygood agreement with the measured C3H8 andH2O profiles at pressures up to 3 bar (Fig. 2).Moreover, the increase in catalytic reactivity withincreasing downstream surface temperatures andthe approach to transport-limited conversion isalso well captured at this pressure range (seeFig. 2(2a) through (2d)).

At pressures above 3 bar, the employed cata-lytic model could not capture the measuredboundary-layer profile of propane (see Fig. 2(3and 4)); therein, the predictions clearly overpre-dicted the measured catalytic reactivity. This isattributed to the overall reaction order (and hencepressure order) of 1.15 (Eq. (1)). Recent catalyticcombustion studies of methane on Pt [18] haveshown that the heterogeneous reactivity of meth-ane follows a �p0.47 pressure dependence, overthe range 1 6 p 6 16 bar; this power law bearsthe combined effects of the first-order reactiondependence with respect to methane, and thep�0.53 dependence due to the reduction of freeplatinum sites with increasing oxygen partial pres-sure. The latter mechanism is very important inrestraining the rate of increase of the catalyticreactivity with rising pressure. The challenge in aglobal catalytic step is to reproduce the reactivityover a wide range of operating conditions (pres-

sures and temperatures). Appropriate pressurecorrections have thus been proposed for globalcatalytic steps of methane [18]; with such correc-tions, the performance of single steps has beenshown to be modest over the wide pressure range1 to 16 bar.

In a fashion similar to the earlier methanestudies, the current global propane step isextended by introducing a pressure term whichrestrains the rate of increase of the catalytic reac-tivity with increasing pressure:

_sC3H8¼ A� p

p0

� ��n

T 1:15w � expð�Ea=RT wÞ � ½C3H8�a;

ð2Þ

with n a positive number smaller than unity, andp0 = 1 bar. A value of n = 0.4 ± 0.03 yielded thebest fit to the experimentally acquired boundarylayer propane profiles over the entire pressurerange 1 6 p 6 7 bar, for all the cases consideredin this work (Fig. 2).

Transverse profiles of major species were com-puted anew for Cases 2–4 with the proposedkinetics of Eq. (2) at locations upstream of consid-erable gas-phase fuel conversion. Significantly bet-ter agreement with the Raman data was achievedfor the higher pressure Cases 3 and 4 (5 and 7 bar,respectively) as seen in Fig. 2, particularly for the

Fig. 3. Computed streamwise catalytic (C) and gas-phase (G) fuel conversion rates, averaged over y (solid,dashed-dotted-lines: pressure-corrected model; dashed,dotted-lines: Garetto). Upper- and lower-wall (solid anddashed gray lines respectively) temperature profiles fittedthrough thermocouple measurements (upper: circles;lower: triangles). The vertical arrows on the x-axesdefine the onset of non-negligible gas-phase conversion(black: pressure-corrected model; gray: Garetto).

Fig. 4. Measured and predicted transverse profiles ofC3H8 and H2O mole fractions for Cases 5, 6, and 7 atthree streamwise locations. Measurements: C3H8 (cir-cles), H2O (triangles). Predictions: C3H8, solid lines;H2O, dashed-dotted lines. For clarity, the scale ofpropane has been expanded by a factor of 2 in Cases 5and 7. Upper- and lower-wall (solid and dotted-lines,respectively) temperature profiles fitted through thermo-couple measurements (upper: open symbols; lower: gray-filled symbols) for Cases 5 (circles), 6 (squares), and 7(triangles).


lower examined surface temperatures. At the sametime, the good model performance at pressuresp 6 3 bar was retained. The large overpredictionof catalytic reactivity by Eq. (1) is evident inFig. 3. For Case 4 (7 bar), the fuel conversion rateis overpredicted by as much as 42% for a surfacetemperature of 786 K, which in turn yields thesame overprediction in heterogeneously-producedwater (Fig. 2(4)). At higher wall temperatures,however, mass-transport-limited operation isapproached and both models yield similar results:the difference between both model predictionsdrops to 5.2% at 995 K. It is noted that theenhanced performance of Eq. (2) yielded a moreconservative estimate of the reactor extent withminimal gas-phase contribution. As seen inFig. 3, the computed lengths of significant gas-phase conversion are reduced by as much as 13and 21 mm for Cases 3 and 4, respectively. Theassociated higher near-wall fuel concentration,along with the increased concentration of catalyt-ically produced water in the gas-phase inductionzone, have been shown to affect the onset ofhomogeneous ignition [19].

The applicability of the pressure-corrected glo-bal step was further evaluated by comparing

numerically predicted species profiles with Ramandata for Cases 5–7 (Fig. 4). The propane andwater profiles are captured correctly at all pres-sures and for all surface temperatures considered.Cases 5 and 7 pertain to propane/air mixtures(without oxygen addition) with same equivalenceratio and mass inflow and similar surface temper-atures, at 3 and 7 bar, respectively. An increasedcatalytic reactivity from 3 to 7 bar is evident bycomparing propane profiles at positionx = 68 mm, with upper wall temperatures of 864and 868 K for Cases 5 and 7, respectively. Case6 is provided to exemplify the aptness of theextended model at lower equivalence ratios (/= 0.23) and lower wall temperatures(723 K 6 Twall 6 898 K).

4.2. Homogeneous ignition

It can be argued that a global catalytic stepthat captures (over a prescribed range of operat-ing parameters) the heterogeneous fuel consump-tion, when used in conjunction with a detailed


homogeneous reaction scheme, can also repro-duce the onset of homogeneous ignition. Earliercatalytic combustion studies of hydrogen andmethane [19,20] with detailed hetero-/homoge-neous reaction schemes have clarified the particu-larly weak coupling between the two pathways viaradical adsorption/desorption reactions; this cou-pling had a minimal effect on the radical pool overthe gas-phase induction zone [19,20,26]. Majorhetero-/homogeneous interactions come from thenear-wall catalytic fuel depletion, which in turninhibits homogeneous ignition [7], and a second-ary chemical coupling originating from the heter-ogeneously produced major species (notablyH2O), which impact homogeneous ignition (pro-moting for methane [17,27], inhibiting for hydro-gen [20,28]). These two effects can be capturedby a global catalytic step. Coupling via otherintermediate species, such as CO, can be impor-tant [16,18] in determining the overall consump-tion of this species and in determining theresulting CO emissions but its impact on homoge-neous ignition of methane is secondary [17]. Previ-ous hydrogen and methane homogeneous ignitionstudies have clearly pointed out that deficienciesin homogeneous ignition predictions stem mostlyfrom low-temperature gas-phase kinetics and notfrom heterogeneous kinetics [17,20].

Comparisons between LIF-measured andnumerically predicted distributions of the OH rad-ical are illustrated in Fig. 5 for Cases 8–11 ofTable 1. In all cases, the pressure-corrected heter-ogeneous model established in the previous sec-tion was used, coupled to the optimized gas-phase mechanism of Qin. The slight asymmetriesof the flames are due to temperature differencesbetween the two catalytic walls. The location ofhomogeneous ignition (xig), shown with the greenarrows in Fig. 5, was determined in both experi-

Fig. 5. Measured and predicted distributions of the OHradical for Cases 8–11. (a) OH-LIF, (b) numericalpredictions with the Qin (gas-phase) and pressure-corrected (catalytic) reaction schemes. The green arrowsdefine the onset of homogeneous ignition. Predicted OHlevels in ppmv.

ments and predictions as the streamwise positionwhere OH levels rose to 5% of their maximumflame values. The OH color-coded bars (ppmv)in Fig. 5 refer to the predictions (the OH-LIFmeasurements were in the linear regime but abso-lute concentrations were not deduced). The lowerpressure flames exhibited the highest absolute OHlevels, which dropped rapidly with increasingpressure.

Having established the applicability of thepressure-corrected global step in the previous sec-tion, the comparisons in Fig. 5 allow for a directevaluation of the gas-phase reaction scheme inpredicting ignition delay times and laminar flamespeeds (linked to the homogeneous ignition dis-tances and to the flame sweep angles inside thechannel, respectively). Figure 5 indicates a moder-ate ability of the gaseous scheme in capturing cor-rectly the measured flame shapes, with betteragreement at lower pressures. However, of greaterimportance is the ability of the gaseous scheme tocorrectly predict homogeneous ignition distances(since homogeneous ignition is deemed as detri-mental to the reactor integrity). An overall goodagreement is achieved between measurementsand predictions; the ignition distance is underpre-dicted by 9.3% for Case 8 (1 bar), while pressuresup to 5 bar consistently yielded overpredictions(by up to 9.8% for Case 10). Increasing the pres-sure to 7 bar resulted in greater overpredictionsof xig (by 19% for Case 11). These observationsare in accordance with recent numerical studiesthat demonstrated the gas-phase mechanism’sinability to correctly capture measured ignitiondelay times of propane at moderate-to-high pres-sures (p > 4 bar) [29].

Further validation for the applicability of thecoupled hetero-/homogeneous models is providedin Fig. 6; Raman-measured major species profilesover the channel half-height are compared against

Fig. 6. Predicted (lines) and measured (symbols) profilesover the channel half-height of C3H8 and H2O molefractions for Case 10 at streamwise positions x = 23(solid lines, diamonds), 53 (dashed-dotted lines, lowertriangles), 83 (dotted lines, squares), 113 (dashed lines,circles), and 143 (dashed-doubled-dotted lines, uppertriangles) mm.

Fig. 7. Predicted catalytic (C) and gas-phase (G) con-version rates for the entire channel length for Case 9,along with propane and OH radical species molefractions. The vertical gray arrow on the x-axis indicatesthe predicted point of homogeneous ignition.


numerical predictions obtained using the pressure-corrected global step and the gas-phase scheme,for selected streamwise positions of Case 10 priorto homogeneous ignition. The predictions are ingood agreement with the experiments, indicatingthat the chemical models capture correctly the fuelconsumption in the reactor.

In Fig. 7, catalytic and gas-phase conversionrates along with propane and OH radical molefractions are plotted for Case 9. It is evident thatgas-phase chemistry is responsible for most of thefuel conversion, well-before the onset of homoge-neous ignition. Specifically, the pre-ignition gas-phase chemistry already surpasses the catalyticchemistry in terms of fuel conversion atx = 11.8 cm, whereby only �32% of the fuel hasbeen converted. The significant amount of fuelavailable for homogeneous reactions is due tothe small surface-to-volume ratio of the channel(dictated by requirements of reactor optical acces-sibility), along with the large Lewis number ofpropane that in turn reduces the fuel transportrate to the catalytic surfaces.

5. Conclusions

The pure heterogeneous and the coupled het-ero-/homogeneous combustion of lean propane/air mixtures over platinum was investigated exper-imentally in an optically accessible, channel flowreactor at pressures 1 bar 6 p 6 7 bar within situ, 1-D Raman and planar OH-LIF measure-ments. The catalytic fuel conversion rate wasassessed as a function of pressure, equivalenceratio and surface temperature. A global-step cata-lytic reaction for propane was established, whichcorrectly accounted for the pressure dependenceof the catalytic reactivity over the entire investi-gated pressure range. Numerical predictions usinga detailed homogeneous reaction scheme coupledwith the established catalytic reaction step repro-

duced the onset of gas-phase ignition at moderatepressures (p 6 5 bar), with somewhat more pro-nounced overpredictions at higher pressures (7bar). The overall performance of the employedhetero-/homogeneous reaction models verifiedtheir applicability for the design of propane-fueledcatalytic microreactors.

Acknowledgments

Support was provided by Paul Scherrer Insti-tute and the Swiss Federal Institute of TechnologyZurich (ETHZ). We are thankful to Mr. R. Scha-eren for the aid in the experiments.

Appendix A. Supplementary data

Supplementary data associated with this articlecan be found, in the online version, atdoi:10.1016/j.proci.2008.06.063.

References

[1] R. Carroni, V. Schmidt, T. Griffin, Catal. Today 75(2002) 287–295.

[2] A. Schneider, J. Mantzaras, R. Bombach, S.Schenker, N. Tylli, P. Jansohn, Proc. Combust.Inst. 31 (2007) 1973–1981.

[3] L.L. Smith, H. Karim, M.J. Castaldi, S. Etemad,W.C. Pfefferle, Catal. Today 117 (2006) 438–446.

[4] J.M. Ahn, C. Eastwood, L. Sitzki, P.D. Ronney,Proc. Combust. Inst. 30 (2005) 2463–2472.

[5] A. Gomez, J.J. Berry, S. Roychoudhury, B. Cori-ton, J. Huth, Proc. Combust. Inst. 31 (2007) 3251–3259.

[6] D.G. Norton, D.G. Vlachos, Proc. Combust. Inst.30 (2005) 2473–2480.

[7] J. Mantzaras, Interplay of transport and hetero-/homogeneous chemistry, in: S.Z. Jiang (Ed.), Focuson Combustion Research, Nova Publishers, NewYork, 2006, p. 241.

[8] W.C. Pfefferle, L.D. Pfefferle, Prog. Energy Com-bust. Sci. 12 (1986) 25–41.

[9] G. Veser, M. Ziauddin, L.D. Schmidt, Catal. Today47 (1999) 219–228.

[10] S.R. Deshmukh, D.G. Vlachos, Combust. Flame149 (2007) 366–383.

[11] M. Aryafar, F. Zaera, Catal. Lett. 48 (1997) 173–183.

[12] O. Deutschmann, L.I. Maier, U. Riedel, A.H.Stroemman, R.W. Dibble, Catal. Today 59 (2000)141–150.

[13] P. Aghalayam, Y.K. Park, N. Fernandes, V.Papavassiliou, A.B. Mhadeshwar, D.G. Vlachos,J. Catal. 213 (2003) 23–38.

[14] T.F. Garetto, E. Rincon, C.R. Apesteguia, Appl.Catal. B Environ. 48 (2004) 167–174.

[15] C. Bruno, P.M. Walsh, D.A. Santavicca, N. Sinha,Y. Yaw, F.V. Bracco, Combust. Sci. Technol. 31(1983) 43–74.

http://dx.doi.org/10.1016/j.proci.2008.06.063


[16] S. Karagiannidis, J. Mantzaras, G. Jackson, K.Boulouchos, Proc. Combust. Inst. 31 (2007) 3309–3317.

[17] M. Reinke, J. Mantzaras, R. Bombach, S. Schen-ker, A. Inauen, Combust. Flame 141 (2005) 448–468.

[18] M. Reinke, J. Mantzaras, R. Schaeren, R. Bom-bach, A. Inauen, S. Schenker, Combust. Flame 136(2004) 217–240.

[19] M. Reinke, J. Mantzaras, R. Schaeren, R. Bom-bach, A. Inauen, S. Schenker, Proc. Combust. Inst.30 (2005) 2519–2527.

[20] C. Appel, J. Mantzaras, R. Schaeren, et al., Com-bust. Flame 128 (2002) 340–368.

[21] Z. Qin, V.V. Lissianski, H. Yang, W.C. Gardiner,S.G. Davis, H. Wang, Proc. Combust. Inst. 28(2000) 1663–1669.

[22] M.E. Coltrin, R.J. Kee, F.M. Rupley, SurfaceChemkin: A Fortran Package for Analyzing Heter-ogeneous Chemical Kinetics at the Solid Surface–GasPhase Interface, Report No. SAND90-8003C, San-dia National Laboratories, 1996.

[23] R.J. Kee, F.M. Rupley, J.A. Miller, Chemkin II: AFortran Chemical Kinetics Package for the Analysisof Gas-Phase Chemical Kinetics, Report No.SAND89-8009B, Sandia National Laboratories,1996.

[24] R.J. Kee, G. Dixon-Lewis, J. Warnatz, M.E.Coltrin, J.A. Miller, A Fortran Computer CodePackage for the Evaluation of Gas-Phase Multicom-ponent Transport Properties, Report No. SAND86-8246, Sandia National Laboratories, 1996.

[25] C. Appel, J. Mantzaras, R. Schaeren, R. Bombach,A. Inauen, Combust. Flame 140 (2005) 70–92.

[26] M. Reinke, J. Mantzaras, R. Schaeren, R. Bom-bach, W. Kreutner, A. Inauen, Proc. Combust. Inst.29 (2002) 1021–1029.

[27] M. Reinke, Mantzaras, R. Bombach, S. Schenker,N. Tylli, K. Boulouchos, Combust. Sci. Technol.179 (2007) 553–600.

[28] P.A. Bui, D.G. Vlachos, P.R. Westmoreland, Proc.Combust. Inst. 26 (1996) 1763–1770.

[29] G. Jomaas, X.L. Zheng, D.L. Zhu, C.K. Law,Proc. Combust. Inst. 30 (2005) 193–200.

Experimental and numerical investigation of the hetero...

Documents

Transcript of Experimental and numerical investigation of the hetero...