lable at ScienceDirect

Journal of Power Sources 274 (2015) 149e155

Contents lists avai

Journal of Power Sources

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

Operating envelope of a short contact time fuel reformer for propanecatalytic partial oxidation

Michael G. Waller, Mark R. Walluk, Thomas A. Trabold*

Golisano Institute for Sustainability and Center for Sustainable Mobility, Rochester Institute of Technology, USA

h i g h l i g h t s

� Simulated and experimental propane cPOx results.� Theoretical efficiency limit is 89%.� Achieved a maximum experimental efficiency of 84%.� Optimal operating point is O2/C ¼ 0.53 at 940 �C.

a r t i c l e i n f o

Article history:Received 31 July 2014Received in revised form19 September 2014Accepted 4 October 2014Available online 13 October 2014

Keywords:MicrolithPropane reformingCatalytic partial oxidationCarbon formationFuel cell

* Corresponding author. 111 Lomb Memorial DriveTel.: þ1 585 475 4696.

E-mail addresses: tatasp@rit.edu, thomas.trabold@

http://dx.doi.org/10.1016/j.jpowsour.2014.10.0250378-7753/© 2014 Elsevier B.V. All rights reserved.

a b s t r a c t

Fuel cell technology has yet to realize widespread deployment, in part because of the hydrogen fuelinfrastructure required for proton exchange membrane systems. One option to overcome this barrier is toproduce hydrogen by reforming propane, which has existing widespread infrastructure, is widely usedby the general public, easily transported, and has a high energy density. The present work combinesthermodynamic modeling of propane catalytic partial oxidation (cPOx) and experimental performance ofa Precision Combustion Inc. (PCI) Microlith® reactor with real-time soot measurement. Much of thereforming research using Microlith-based reactors has focused on fuels such as natural gas, JP-8, diesel,and gasoline, but little research on propane reforming with Microlith-based catalysts can be found inliterature. The aim of this study was to determine the optimal operating parameters for the reformer thatmaximizes efficiency and minimizes solid carbon formation. The primary parameters evaluated werereformate composition, carbon concentration in the effluent, and reforming efficiency as a function ofcatalyst temperature and O2/C ratio. Including the lower heating values for product hydrogen and carbonmonoxide, efficiency of 84% was achieved at an O2/C ratio of 0.53 and a catalyst temperature of 940 �C,resulting in near equilibrium performance. Significant solid carbon formation was observed at muchlower catalyst temperatures, and carbon concentration in the effluent was determined to have a negativelinear relationship at T < 750 �C. The Microlith reactor displayed good stability during more than 80experiments with temperature cycling from 360 to 1050 �C.

© 2014 Elsevier B.V. All rights reserved.

1. Introduction

Over the past few decades, fuel cells have gained traction in theautomotive and large-scale power generation industries, but haveyet to gain a foothold in the residential sector. While there havebeen several fuel cell systems developed for residential use, there isa gap in systems that can deliver power on the order of250 We1 kW. Some examples of residential devices that require

, Rochester, NY 14623, USA.

gmail.com (T.A. Trabold).

these power levels include lawn mowers, leaf blowers, gas trim-mers, and backup auxiliary power units. Presently, residentialsystems that require electrical power at this mid-level are domi-nated by small internal combustion engines (ICE) that suffer fromlow electrical conversion efficiency, high emissions, and loudoperation. This provides a unique opportunity for fuel cell tech-nology as it directly addresses these major drawbacks of ICE-baseddevices.

In spite of their myriad advantages, fuel cell systems capable ofmeeting medium-sized power requirements for the residentialsector have yet to be commercialized at high volume, in partbecause of issues concerning the fuel quality requirements andattendant supply infrastructure. Hydrogen is predominately

M.G. Waller et al. / Journal of Power Sources 274 (2015) 149e155150

present in compound form and must first be processed, through amethod such as hydrocarbon reforming, before it can be used as anenergy carrier. Therefore, a key component of a viable fuel cellpower generation system is that its fuel supply must be readilyavailable, acceptable to regulatory/safety agencies, and conven-tional for the general public's acceptance. Propane, is an attractivefuel because it already has widespread infrastructure in place, thegeneral public is familiar with its use, it is transportable, and has ahigh energy density. The Golisano Institute for Sustainability (GIS)at Rochester Institute of Technology (RIT) is developing a workingfuel cell system that produces approximately 500 W, and runs oncommercial propane. A general outline for the program involvesthree phases including the reformation system design, fuel cellstack configuration, and the final system integration. This workfocuses on the first phase, fuel reformation.

Common catalytic reformation techniques include steamreforming (SR), catalytic partial oxidation, dry reforming (DR), andauto-thermal reforming (ATR). It is well known that SR is the mostwidely used process for hydrogen production at very large scales,however, SR may not be best suited for small scale portable ap-plications due to additional system components such as an addi-tional water tank and water management system [1]. Partialoxidation is a valid alternative since the process only requires fueland air as oxidant, and is an exothermic reaction. However, whenair is used as an oxidant, the products are dilutedwith nitrogen. Themechanism for cPOx has been investigated in many studiesincluding [2e4] and is an exothermic reaction with CO and H2 asthe primary reaction products. Primary catalyst degradation modesinclude metal sintering from excessive thermal stress, coke depo-sition, and sulfur contamination [5e7].

While reformation processes have already been developed atvery large scales with high efficiencies, scaling down the reformerfor small mobile applications creates significant challenges due tothe requirement for quick startup and transient response, a smalloperating volume, and high efficiencies. The Short Contact Time(SCT) reactor design is an approach that allows for high volumes offluid flow such that the residence time of the gas mixture inside thecatalyst bed is on the order of milliseconds [8]. Much of theresearch using SCT reactors has been focused on syngas/hydrogenproduction [9], however other applications include dehydrogena-tion [10e12], and Fischer-Tropsch synthesis [13]. Some of the morecommonly studied SCT reactors include ceramic monoliths andfoams [14e16].

The reformer studied in this paper was provided by PrecisionCombustion Inc. (PCI), and is based on their patented Microlith®

technology [17]. The Microlith substrate consists of a series ofcatalytically coated metal meshes with very small channel di-ameters, allowing for ultra-short-channel-length. As compared toother SCT substrates such as ceramic monoliths and foams, theMicrolith reactor provides a higher rate of reaction, lower thermalmass, and significantly higher heat and mass transfer coefficients[18]. The Microlith substrate performs favorably because of its highgeometric surface area (GSA) per unit volume of the reactor, andhigh specific surface area of the catalyst support/washcoat [19].Additionally, the low thermal mass of the Microlith elements al-lows for rapid start-up and transition to steady operating temper-atures. Using a lumped sum capacitance model, a 30-fold increasein thermal response time was determined when comparing aMicrolith reactor to a ceramic monolith substrate [20]. Further-more, near theoretical equilibrium reactions for partial oxidationexperiments have been observed using these reactor designs whichresults in high selectivity of the desired product as well as highreformer efficiencies [19].

Fuel reforming research using Microlith based catalysts havefocused on fuels including gasoline [8], methane [21], methanol

[22], JP-8 and diesel [23], but little research on propane reformingwith a Microlith catalyst can be found in literature. The aim of thisarticle is to assess the reforming performance of a 31.75 mmdiameter Inconel single tube reactor for propane cPOx, and toexamine various operating points. Previous collaborativeworkwithPCI has yielded impressive results using their Microlith catalyststhat were developed for syngas production [18]. The operatingenvelope for this reformer is described in terms of temperaturerange, and the diatomic oxygen (from air)-to-carbon ratio (O2/C).The catalytic performance is evaluated in terms of H2 and CO yield,conversion efficiency, and solid carbon formation. Additionally, theoptimal operating point that maximizes fuel conversion efficiency,and minimizes carbon formation is described. Thermodynamicmodeling of propane cPOx was conducted using the Gibbs freeenergy minimization method with aspenOne® Engineering soft-ware, to complement the experimental data. All experimentaltesting was performed at an existing fuel cell and bio-fuel researchfacility at RIT's Golisano Institute for Sustainability (GIS).

2. Methods and materials

2.1. Thermodynamic modeling

In order to simulate the propane cPOx reaction, aspenOne®

Engineering modeling software was employed. AspenOne Engi-neering is a comprehensive chemical process modeling softwareand is widely used in the petrochemical industry for refiningsimulation and optimization [24]. The propane cPOx model simu-lation was developed using Aspen HYSYS V8.4 of the aspenOneEngineering package, and primarily utilized the Gibbs reactor unitoperation block. The Soave-Redlich-Kwong (SRK) equation of statewas the global property model used. In order to simulate the effectsof temperature and input O2/C ratio on propane cPOx, a case studywas developed that varied the reformer temperature and O2/C ratiowith a propane fuel flow of 1 kmol s�1. Previous investigations ofpropane reforming generally operate catalysts between 400 and800 �C [4,25,26]. For the simulated case study, the reformer tem-perature was varied from 400 to 1000 �C with a step size of 10 �C.

For the propane cPOx simulation, the following assumptionswere made:

1) Uniform temperature distribution within the reformer.2) The reformate gas mixture behaves as an ideal gas and pres-

sure gradients were ignored within the reactor.3) Inlet propane and air were well mixed and the reaction rea-

ches thermodynamic equilibrium. The considered reformatespecies in this analysis were H2, CO, O2, H2O, CH4, C2H4, C2H6,CO2, C3H8, C (solid graphite), and N2.

4) The Gibbs reactor unit operates adiabatically.5) The major modeling independent parameters were reformer

temperature and O2/C ratio. The effect of GHSV was notconsidered for this model.

The stoichiometric equilibrium for propane cPOx occurs at anO2/C ratio of 0.5 as can be seen in Equation (1) [27].

C3H8 þ 1:5O2/4H2 þ 3CO; DH298�C ¼ �229 kJ=mol (1)

Thus the simulated reaction was designed to vary the O2/C ratiofrom 0.3 to 1.2 by changing the oxygen inlet flow from 1 to3.5 kmol s�1 with a step size of 0.1. Results are plotted herein fromthis analysis using Matlab®, including the effluent composition andreformer efficiency as a function of temperature and O2/C ratio. Thereformer efficiency was calculated based on the followingequation:

M.G. Waller et al. / Journal of Power Sources 274 (2015) 149e155 151

href ¼_xH2

�LHVH2þ _xCO�LHVCO

_xC3H8�LHVC3H8

(2)

where _x represents the flow rate of the various gases in g s�1, andLHV represents the lower heating values that are 120.1 kJ g�1,10.1 kJ g�1, and 46.4 kJ g�1 for hydrogen, carbon monoxide, andpropane respectively [27]. From the results of this simulation, theoptimal operating conditions that maximize efficiency and mini-mize carbon formation are proposed and compared to experi-mental results.

2.2. Experimental methods and materials

Fig.1 illustrates a schematic and the actual setup for the propanecPOx experiments. All tests were performed on a 31.75 mm diam-eter Inconel reactor tube containing PCI's Microlith based catalyst.The reformer was constructed and delivered to RIT by PCI with asingle thermocouple installed directly on to the catalyst. Thecatalyst itself takes up only a small portion in the center of thereactor tube, and is 8.9 mm in length. The reformer was entirelyencased within an electric tube furnace and was heated to obtainthe desired catalyst temperatures for light-off. Once the light-offtemperature was reached, upstream propane and air were sentthrough the reactor. Influent gas flows were controlled using Alicatmass flow controllers specifically calibrated to control propane andair flows within ±1%. The propane used for these tests was indus-trial grade supplied by Airgas with a sulfur content of <15 ppmv.The effluent gas species were measured using a mass spectrometer(Applied Instrument Technologies, CA, USA), gas chromatograph(Model 490 Micro GC, Agilent Technologies, Santa Clara, CA) thatwere calibrated to quantify the species H2, CO, O2, H2O, CH4, C2H4,C2H6, CO2, C, and N2. Additionally, an AVL micro-soot meter (ModelNo. 483, AVL List GmbH, Graz, Austria) was used to measure thesolid carbon formation of the reformate stream. This instrumentutilizes the photo-acoustic effect to obtain accurate carbon con-centration measurements as low as 5 mg/m3.

The input and effluent compositions were all recorded using areforming analysis programwritten in LabView. Each test point wasrun for a minimum of 20 min with the average value over this timeperiod recorded.

The test apparatus employed the reformer in a vertical orien-tation with the propane and air introduced from the bottom. Theinlet streams were not pre-heated and they were mixed with acustommanufactured tee manifold. The manifold was insulated upto the furnace and thermocouples measured the temperature of the

Fig. 1. Left: fuel reformer facility with associated instrumentation, right: propane cPOx testcoated meshes) catalyst technology (31.75 mm diameter, 8.9 mm length).

inlet stream. Thermocouples were also installed upstream, internalto, and downstream of the catalyst. The catalyst thermocouple wasinstalled by PCI, and was placed within the catalyst bed. Inlet andreformer differential pressure transducers were integrated into thestand design.

A multi-parameter experimental matrix including O2/C, catalysttemperature, and gas hourly space velocity was developed for thepropane cPOx reformation experiments. The propane flow rateused for all the experiments (0.02 g s�1) was based on the esti-mated flow required to produce 500Wof power after reformer andfuel cell efficiency losses. Except for the light-off temperaturedetermination tests, the general method for reforming experimentswas to preheat the catalyst up to a minimum of 10 �C above theknown light off temperature, whereupon propane was thenallowed to flow. Air was then sent through the reformer andadjusted until the desired O2/C was obtained. It can be noted thatonce propane and air were sent through the catalyst, the catalysttemperature would spike to the 700e800 �C range in a matter ofseconds. Based on the stoichiometric equilibrium, Equation (1), anO2/C ratio of greater than 0.5 was used for all experiments. Thespace velocity was determined using the void volume of the cata-lyst and the flow of the reactants at standard temperature andpressure. The gas hourly space velocity (GHSV) was controlledthrough a range of 921 hr�1 to 1242 hr�1.

The initial tests explored the temperature boundaries of thereformer including the light-off temperature and the auto-ignitiontemperature. The remainder of test points was conducted at oper-ating conditions that would not cause irreversible catalyst degra-dation by maintaining the catalyst temperatures below 1000 �C.The majority of test points were taken by first sending the properamount of air flow as determined by the O2/C ratio, then the tubefurnace set point temperature was adjusted until the desiredcatalyst temperature was reached. For each O2/C ratio, the catalysttemperature was increased from 800 �C to 1000 �C with a step sizeof 20 �C. Data was collected at each test point for a minimum of20min. For reporting, the average value over the 20min period wasused. The reformer efficiency was calculated based on Equation (2).

3. Results and discussion

3.1. Thermodynamic simulation results

From Equation (1), it can be seen that complete propane con-version cannot occur below an O2/C ratio of 0.5. However,increasing the O2/C ratio beyond 0.5may lead to increased amounts

cell schematic containing Inconel single tube reactor based on Microlith (catalytically

M.G. Waller et al. / Journal of Power Sources 274 (2015) 149e155152

of undesired by-products such as CO2 and H2O in the effluent gas.To develop an operating test matrix, a Gibbs free energy analysiswas conducted using aspenOne Engineering software by estimatingthe effluent gas composition that results from propane cPOx underequilibrium conditions. Fig. 2 displays the effluent compositionsfrom this analysis for H2, CO, CH4, CO2, C, and the efficiency of thereaction.We assumed a uniform equilibrium temperature along thereactor for the analysis which may not be likely in reality. However,because experimental catalyst temperature data were measuredfrom a thermocouple placed directly on the Microlith catalyst, weexpect the equilibrium temperature used in the Gibbs analysis tocorrespond well to experimental methods. The H2 yield, CO yield,and efficiency plots all follow similar trends where the maximumproduction points occur at O2/C ¼ 0.53, and a temperature of1000 �C. Conversely, the unwanted products such as CH4 and CO2have high production rates at high O2/C ratios and temperaturesaround 700 �C. For O2/C ratios below 0.53, hydrogen productiondecreases rapidly regardless of temperature however at ratiosabove 0.53, the H2 production does not drop off as quickly. The COand efficiency plots follow similar trends. The carbon free region

Fig. 2. Equilibrium analysis of propane cPOx effluent composition yield and reformer effireversed in 2c to enable adequate observation of surface plots).

occurs at temperatures above 650 �C with O2/C ratios of less than0.6.

Fig. 3 shows the expected specific compositions of effluent gasesat an O2/C ratio of 0.53 as a function of temperature. Once the re-action reaches a temperature of approximately 800 �C, increasingthe temperature further does not have much of an impact on thereaction efficiency and effluent gas composition. Additionally, atabove 850 �C, essentially no CH4, C2H4, C2H6, and CO2 are producedin this reaction and the effluent gas is composed almost entirely ofH2, CO, and inert N2.

Similarly with solid carbon formation, when operating the re-action under equilibrium conditions and at temperatures above750 �C, there is little to no coke formation. This is most likely causedby the Boudouard reaction which is the disproportionation of COinto CO2 and solid carbon as seen in Equation (3) [28].

2CO#CO2 þ C (3)

At high temperatures, the reverse Boudouard reaction isfavored, converting CO2 and C into CO while at low temperatures

ciency as a function of O2/C and catalyst temperature. (Note: scale orientations were

Fig. 3. Anticipated effluent gas composition as a function of temperature at O2/C ¼ 0.53.

M.G. Waller et al. / Journal of Power Sources 274 (2015) 149e155 153

CO is disproportioned into CO2 and C. The effects of this on thereformate constituents can be seen in both the carbon dioxide andsolid carbon (Fig. 2d and e, respectively). As the temperature of thereaction increases, both carbon dioxide and solid carbon formationdecrease. Thus tomitigate coke formation, the reactionmust be runat a sufficiently high temperature. From Fig. 2e, it appears thattemperatures above 750 �C are sufficient to limit solid carbonformation.

Based on the thermodynamic simulation, it is anticipated thatthe optimal operating point for the reformer should lie between800 and 1000 �C with an O2/C z 0.53. The theoretical maximumefficiency of the reformer under equilibrium conditions can becalculated using Equation (2), and is approximately 89% as theeffluent gases are composed almost entirely of H2 and CO with theinert N2. Under these conditions, it is anticipated that little to nocoke formation will occur. Several experiments within this tem-perature range and approximate O2/C ratio were conducted tovalidate these results, and are discussed in the following sections.

3.2. Experimental results

3.2.1. Propane auto-ignition temperatureThe Gibbs free energy analysis reveals that the optimal oper-

ating point for propane cPOx should lie within the temperaturerange of 800e1000 �C, with an O2/C ofz0.53. Experimental testingwas conducted to corroborate the Gibbs energy analysis. Table 1lists some of the primary test matrix points to assess the PCIreactor for propane cPOx reforming.

There are two components of the reformer that require tem-perature control: the inlet fuel mixture and the reformer catalyst.The temperature of the fuel and air inlet mixture must be regulatedbecause if the mixture becomes too hot, it will ignite before itreaches the catalyst bed. It is important to understand this valuebecause premature ignition could have harmful impacts on thesystem and its users. One study described propane auto-ignitiontemperatures at around 577 �C [29]. A test run to determine theauto-ignition temperature of propane in the reformer was con-ducted by metering a small amount of propane while steadily

Table 1Experimental test points.

Testpoint

O2/C Propane flow(g s�1)

Air flow(g s�1)

Temperaturerange (�C)

Temperaturestep size (�C)

TP1 0.53 0.02 0.100 800e1000 20TP2 0.55 0.02 0.103 800e1000 20TP3 0.57 0.02 0.106 800e1000 20

increasing the furnace temperature, thus increasing temperature ofthe inlet gas to the reformer. The results in Fig. 4 show the inlettemperature of the reformer, measuring the temperature of the fuelmixture right before it enters the catalyst. At approximately 557 �C,the trumpet temperature began to increase rapidly, revealing thepropane auto-ignition temperature had been reached.

3.2.2. Light-off temperaturePrevious studies investigating propane reforming have found

light-off temperatures ranging between 200 and 370 �C dependingon the catalyst type [4,30,31]. The light-off temperature determi-nation tests for this study were conducted by pre-heating thecatalyst to around 300 �C, then metering the fuel-air mixturethrough the reformer. The furnace set point was then increased,thus increasing the catalyst temperature. Due to the exothermicnature of the cPOx reaction, the temperature of the catalyst in-creases rapidly once propane conversion begins indicating thelight-off temperature. The light-off temperature was determined tobe approximately 360 �C.

3.2.3. Temperature operating limits and coke formationThe maximum operating temperature is limited by the pro-

pensity for metal sintering. In this experimental campaign, thereformer temperature was kept below 1050 �C in order to maintainthe longevity of the catalyst. Tests where the catalyst was at atemperature below 1050 �C were presumably safe, however, if thetemperature drops too low there is a risk of solid carbon formationand subsequent build-up inside the catalyst and/or downstreamfuel cell, causing long-term degradation. Thus, the low operatingtemperature point is dictated by the point where significant carbonformation begins to occur, most likely due to the Boudouard reac-tion. From Fig. 2e, the carbon free zone occurs at O2/C ratios below0.6 and temperatures above 750 �C. At an O2/C¼ 0.53 and using theAVL micro-soot meter, the amount of solid carbon was measuredafter taking samples of the effluent gas. Fig. 5 shows an inverserelationship between carbon concentration in the effluent streamand catalyst temperature for experimental data.

According to the experimental data, coke formation is stronglycorrelated to catalyst temperature. Significant carbon formationbegan at approximately 750 �C, and steadily increased as thecatalyst temperature was decreased. Above 800 �C, very little, ifany, solid carbon was formed. To operate the reformer conserva-tively and to avoid known degradation modes, nearly all of theremaining tests were conducted with the catalyst operating withinthe temperature range of 800e1000 �C.

3.2.4. Effects of temperature and O2/C ratioFrom the thermodynamic simulation, the most efficient oper-

ating point for propane cPOx exists within the temperature range of

Fig. 4. Inlet temperature of the reformer consisting of the Microlith based catalyst.

Fig. 5. Experimental soot (solid carbon) formation from propane cPOx as a function ofcatalyst temperature operating with O2/C ¼ 0.53. Fig. 7. Propane cPOx yield and reformer efficiency at O2/C ¼ 0.57.

M.G. Waller et al. / Journal of Power Sources 274 (2015) 149e155154

800e1000 �C, and an O2/C ratio from 0.5 to 0.6. The main methodfor running the experiments was to set the O2/C ratio and stepthrough the temperature range by increasing the furnace setpointand thus increasing the effective catalyst temperature. Due tosafety concerns, test points for catalyst temperatures around1000 �C could not be obtained because the trumpet temperature ofthe test stand approached the propane auto-ignition temperatureof 560 �C. However, based on the Gibbs free energy analysis, weanticipate little increase in efficiency with catalyst temperaturesabove 950 �C. The GHSV was held within the range of 921 hr�1 to1242 hr�1. Overall, minor deviations in the GHSV had little impacton reactor performance. Figs. 6 and 7 show the results of two testsperformed at O2/C ratios of 0.53 and 0.57. Identifying the effect of awide range of GHSV on the catalyst performance was beyond thescope of this effort.

The results shown in Fig. 6 display themost promising outcomesfrom our tests, indicating that a reformer efficiency of 84% wasachieved while operating the catalyst at a temperature of 940 �Cwith O2/C of 0.53. This efficiency point is 94% of the theoretical limitbased on Equation (2), obtained from our thermodynamic analysis.While the maximum efficiency is obtained at temperaturesapproaching 1000 �C, temperatures down to 800 �C had littlenegative impact on hydrogen yield. Fig. 7 shows the results oftesting at an O2/C of 0.57 where an efficiency of 81% was achievedrevealing some flexibility in the air to fuel ratio. The tabulated datafor both of these tests can be found in Table 2. It is worth noting thatthe maximum performance was observed during the latter part ofthe 130 h of recorded test time. Additionally, we cycled the reactorat least 80 times during these tests, which to some degree can beviewed as simulated startestop cycles. Startup times for the

Fig. 6. Propane cPOx yield and reformer efficiency at O2/C ¼ 0.53.

reformer have not yet been explicitly tested and optimized; how-ever, it is anticipated that once the reformer reaches its light-offtemperature, the catalyst can reach the optimal operating tem-perature in less than 30 s.

It should be noted that in a real-world application, control of thetemperature of the catalyst may not be as simple as in these ex-periments. As more air is added to the reactor, the generated heatfrom the reaction increases, thus driving up the catalyst tempera-ture. However, additional air into the reactor further reduces theamount of H2 generated from the reaction, and therefore the O2/Cratio must be kept at the optimal point (0.53) for maximum per-formance. To determine the relationship between O2/C and thecatalyst temperature at the optimal O2/C ratio of 0.53, several testswere conducted tomeasure the catalyst temperaturewithout aid ofthe furnace. Thiswas completed by simply opening the furnace doorand removing the insulation surrounding the reformer once partialoxidation began. After the furnace had cooled to below 100 �C, thecatalyst temperature would settle around 820 �C. This resultrevealed that for optimal operation of a real system, additionalinsulation or additional heat from some source (possibly externalcombustion of propane) for the reactor is desired. However, even atan operating temperature of 820 �C, hydrogen molar yield wasapproximately 28.0% as compared to 29.4% at a temperature of940 �C, indicating some temperature flexibility in the operatingprotocol. Future work integrating the reformer with a fuel cell stackwill focus onquick startupprocedures as the catalystmust beheatedto approximately 360 �C before fuel reforming can begin.

4. Conclusion

Based on the reforming experiments performed and reportedhere, propane is an attractive primary fuel for small scale resi-dential fuel cell systems due to ease of reforming, transportability,familiarity, and high energy density. Of the various fuel reformingreactions, partial oxidation is advantageous for a small scale resi-dential system because the only reactants are fuel and air, thusgreatly simplifying the system design. In this work, the operationalenvelope and the optimal operating point for PCI's Microlith reactorunder propane cPOx conditions were determined. Simulated re-sults suggested that the most efficient operating point for maxi-mizing hydrogen yield and reformer efficiency exists at catalysttemperatures between 800 and 1000 �C and an O2/C ratio from 0.5to 0.6. Within this range, it is anticipated that little to no cokeformation will occur. Experimental tests revealed a maximum 84%reforming energy efficiency at a temperature of 940 �C and O2/C ¼ 0.53. Little to no solid carbon formation occurred at thisoperating point however, significant carbon formation was

Table 2Experimental propane cPOx data.

O2/C Catalysttemperature (�C)

Reformer input (g s�1) Reformer yield (mol %)

Air Propane H2 CO N2 CH4 CO2 Net syngas (H2 þ CO)

0.53 841 0.100 0.020 28.10 23.32 47.30 0.28 0.94 51.420.53 879 0.100 0.020 28.40 23.58 47.00 0.07 0.76 51.980.53 940 0.100 0.020 29.43 22.82 46.78 0.07 0.87 52.250.57 840 0.106 0.020 26.30 20.39 50.14 0.59 2.54 46.690.57 920 0.106 0.020 26.85 21.54 49.55 0.18 1.86 48.380.57 998 0.106 0.020 27.89 21.40 49.08 0.05 1.55 49.29

M.G. Waller et al. / Journal of Power Sources 274 (2015) 149e155 155

observed at catalyst temperatures below 750 �C. The light-offtemperature for this reformer was found to be approximately360 �C. Future work in this project will be focused on fuel cell andreformer integration. Integrating the reformer with a fuel cell is nota trivial task and we anticipate there being several challenges suchas thermal management, system startup, and minimization ofparasitic losses.

Acknowledgment

This work was performed under the financial assistance ofaward #60NANB13D71 from U.S. Department of Commerce, Na-tional Institute of Standards and Technology, which provided aGraduate Research Assistantship for Michael G. Waller. Dr.Muhammad Arif of the NIST Center for Neutron Research isacknowledged for supporting this effort, and for his group'songoing support of research and development advancing fuel celltechnology in the U.S. We would also like to thank Drs. ChristianJunaedi and Subir Roychoudhury of Precision Combustion Inc. forsupplying the Microlith catalyst and for their technical support.

References

[1] S. Ayabe, H. Omoto, T. Utaka, R. Kikuchi, K. Sasaki, Y. Teraoka, K. Eguchi, Appl.Catal. Gen. 241 (1e2) (Feb. 2003) 261e269.

[2] V. Recupero, L. Pino, R. Di Leonardo, M. Lagana', G. Maggio, J. Power Sources 71(1e2) (Mar. 1998) 208e214.

[3] B. Christian Enger, R. Lødeng, A. Holmen, Appl. Catal. Gen. 346 (1e2) (Aug.2008) 1e27.

[4] P. Corbo, F. Migliardini, Int. J. Hydrogen Energy 32 (1) (Jan. 2007) 55e66.[5] T.A. Trabold, J.S. Lylak, M.R. Walluk, J.F. Lin, D.R. Troiani, Int. J. Hydrogen En-

ergy 37 (6) (Mar. 2012) 5190e5201.[6] H. Pennemann, V. Hessel, G. Kolb, H. L€owe, R. Zapf, Chem. Eng. J. 135 (Suppl. 1)

(Jan. 2008) S66eS73.[7] G. Zeng, Y. Tian, Y. Li, Int. J. Hydrogen Energy 35 (13) (Jul. 2010) 6726e6737.[8] M. Castaldi, M. Lyubovsky, R. LaPierre, W.C. Pfefferle, S. Roychoudhury, SAE

Int. (2003), 2003-01-1366.

[9] A.S. Bodke, S.S. Bharadwaj, L.D. Schmidt, J. Catal. 179 (1) (Oct. 1998) 138e149.[10] M. Huff, L.D. Schmidt, J. Phys. Chem. 97 (45) (Nov. 1993) 11815e11822.[11] M. Fathi, R. Lødeng, E.S. Nilsen, B. Silberova, A. Holmen, Catal. Today 64 (1e2)

(Jan. 2001) 113e120.[12] B. Silberova, M. Fathi, A. Holmen, Appl. Catal. Gen. 276 (1e2) (Nov. 2004)

17e28.[13] R.M. de Deugd, F. Kapteijn, J.A. Moulijn, Catal. Today 79e80 (Apr. 2003)

495e501.[14] I. Aartun, B. Silberova, H. Venvik, P. Pfeifer, O. G€orke, K. Schubert, A. Holmen,

Catal. Today 105 (3e4) (Aug. 2005) 469e478.[15] B.E. Traxel, K.L. Hohn, Appl. Catal. Gen. 244 (1) (May 2003) 129e140.[16] B. Silberova, H.J. Venvik, A. Holmen, Catal. Today 99 (1e2) (Jan. 2005) 69e76.[17] W.C. Pfefferle. Microlith catalytic reaction system, US5051241 A, 24-Sep-1991.[18] C. Junaedi, S.A. Vilekar, D. Walsh, R. Mastanduno, C. Morgan, S. Roychoudhury,

Int. J. Hydrogen Energy 37 (13) (Jul. 2012) 10435e10443.[19] S. Roychoudhury, M. Castaldi, M. Lyubovsky, R. LaPierre, S. Ahmed, J. Power

Sources 152 (Dec. 2005) 75e86.[20] R.N. Carter, S. Roychoudhury, G. Muench, H. Karim, W. Pfefferle, MRS Online

Proc. Libr. 454 (1996).[21] M. Lyubovsky, S. Roychoudhury, R. LaPierre, Catal. Lett. 99 (3e4) (Feb. 2005)

113e117.[22] M. Lyubovsky, S. Roychoudhury, Appl. Catal. B Environ. 54 (4) (Dec. 2004)

203e215.[23] S. Roychoudhury, M. Lyubovsky, D. Walsh, D. Chu, E. Kallio, J. Power Sources

160 (1) (Sep. 2006) 510e513.[24] “AspenTech: Optimizing Process Manufacturing,” Optimize Design and Op-

erations with AspenONE Engineering, 2014. Available from: http://www.aspentech.com/products/aspenone-engineering/ (accessed 18.02.14).

[25] V. Modafferi, G. Panzera, V. Baglio, F. Frusteri, P.L. Antonucci, Appl. Catal. Gen.334 (1e2) (Jan. 2008) 1e9.

[26] C. Resini, M.C. Herrera Delgado, L. Arrighi, L.J. Alemany, R. Marazza, G. Busca,Catal. Commun. 6 (7) (Jul. 2005) 441e445.

[27] NIST, Reference Fluid and Thermodynamic and Transport Properties Database(REFPROP): Version 9.1, in: NIST Standard Reference Database 23, 2013.Available from: http://www.nist.gov/srd/nist23.cfm (accessed 29.07.13).

[28] J.-W. Snoeck, G.F. Froment, M. Fowles, Ind. Eng. Chem. Res. 41 (17) (Jul. 2002)4252e4265.

[29] P. Cadman, G.O. Thomas, P. Butler, Phys. Chem. Chem. Phys. 2 (23) (Jan. 2000)5411e5419.

[30] B.S. Ça�glayan, A.K. Avcı, Z._I. €Onsan, A.E. Aksoylu, Appl. Catal. Gen. 280 (2) (Mar.2005) 181e188.

[31] A.K. Avcı, D.L. Trimm, A.E. Aksoylu, Z._I. €Onsan, Catal. Lett. 88 (1e2) (May 2003)17e22.