- Email: [email protected]
Hydrogen production from rich combustion in porous media
International Journal of Hydrogen Energy 30 (2005) 579 – 592
www.elsevier.com/locate/ijhydene
Hydrogen production from rich combustion in porous media H. Pedersen-Mjaanes, L. Chan, E. Mastorakos∗ Department of Engineering, University of Cambridge, Cambridge, CB2 1PZ, UK Accepted 5 May 2004 Available online 15 July 2004
Abstract This paper examines rich combustion of methanol, methane, octane and automotive-grade petrol inside inert porous media in an e2ort to examine the suitability of the concept for hydrogen production. Species concentrations were measured and operating limits were tested of steady rich 4ames stabilized inside a two-layer alumina foam burner and a two-layer alumina bead burner. Using a conversion e7ciency based on lower heating values, up to 56% of the methanol was converted to syngas (H2 , CO) inside the alumina foam burner and 66% inside the alumina bead burner. Using the same e7ciency de;nition, 45% percent of the methane and 36% of the octane and petrol was converted to syngas with a signi;cant portion of the energy remaining trapped in CH4 , C2 H2 and C2 H4 . For methanol, the highest equivalence ratio observed for stable combustion was 6.1 inside the foam burner and 9.3 inside the bead burner which are higher than the conventional upper 4ammability limit (UFL) of 4.1. Methane’s UFL was increased to 1.9 and, at a minimum, the conventional upper 4ammability limits of iso-octane and petrol were attained. A wide operating envelope was observed, which allowed for large turndown ratios up to 20:1. The composition of the products of the methanol 4ames examined here were close to equilibrium for relatively low equivalence ratios, while those of hydrocarbon 4ames di2ered signi;cantly from equilibrium for all suggesting that ;nite rate kinetics are important. The high conversion e7ciencies, quick startup times, compact size, and the absence of a catalyst make the present burner suitable for consideration as part of a reformer in a fuel cell powered automobile. ? 2004 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved. Keywords: Porous media burner; Hydrogen; Rich combustion; Partial oxidation
1. Introduction Hydrogen contains more energy per unit mass than any other fuel and produces minimum emissions when combusted and essentially no emissions when electrochemically converted to electricity in a fuel cell. It has the potential to reduce dependence on imported fuels depending on how it is generated. These are a few of the reasons why hydrogen is considered the fuel of the future. However, making the transition to a hydrogen economy means converting the existing fossil fuel infrastructure that is already developed. An alternative is to use a fuel reformer to extract hydrogen from existing fuels and/or alcohols. For example, ∗
Corresponding author. Tel.: +44-1223-332690; fax: +441223-332065. E-mail addresses: [email protected] (H. Pedersen-Mjaanes), [email protected] (E. Mastorakos).
an automobile could be powered by any number of fuels currently available, including methanol, ethanol, gasoline and even diesel; they could operate more e7ciently and produce fewer emissions, at the same time creating a bridge between the future hydrogen economy and the present fossil fuel economy. Although numerous technologies exist for producing hydrogen, the three main methods are steam reforming, autothermal reforming and partial oxidation (rich combustion). Partial oxidation and autothermal reforming processes have better dynamic responses and are more compact than a comparative steam reformer [1–3]. Usually, they involve a catalyst which ensures good conversion, but a catalyst is prone to poisoning and requires great care with the sulphur content of the fuel and with particulates. One of the most promising non-catalytic partial oxidation techniques is combustion inside a porous medium. This medium could be a 4uidized or a stationary bed. In a
0360-3199/$ 30.00 ? 2004 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.ijhydene.2004.05.006
580
H. Pedersen-Mjaanes et al. / International Journal of Hydrogen Energy 30 (2005) 579 – 592
Nomenclature A CEA D dm k L LHV m˙ P Pe PPC
cross sectional area (m2 ) Chemical Equilibrium with Applications diameter (m) characteristic pore size (m) thermal conductivity (W=m K) length (m) lower heating value (kJ/kg) mass 4ow rate (kg/s) pressure (kPa) PLeclet number pores per centimetre
4uidized bed system, the incoming velocity of the gases is high enough to lift the particles. A spouted-bed combustor involves motion of solid inert (or catalytic) particles in a recirculatory manner, this technique has been used to stabilise rich 4ames of methane in order to produce syngas [4]. Combustion in stationary porous media occurs inside the voids of the porous matrix. Energy is transferred to the solid by convection from the hot gases. Conduction and radiation through the solid porous medium, which has much higher thermal conductivity than the gas mixture and additionally radiates signi;cantly, distributes some of this energy to the region upstream of the 4ame. This in turn transfers energy via convection to the incoming reactants. The porous matrix acts in essence as an integral preheater. This preheating allows for higher combustion temperatures and increased stability compared to a free premixed laminar 4ame. Locally, the 4ame zone may reach temperatures higher than the adiabatic 4ame temperature of the unburnt mixture, which gives the name “superadiabatic” to this combustion technology [5–10]. The same e2ect is also produced by the recirculation of solid particles in the spouted-bed combustor of Weinberg et al. [4]. Filtration combustion inside a stationary, inert porous medium has also been used to produce syngas [8–10]. In this technique, a con;ned wave was allowed to propagate along the solid, controlled by periodically reversing of the reactants 4ow. Currently, several gaseous fuels and kerosene have been tested using this method. A disadvantage of this technique compared to a stationary 4ame system may be its relatively large size and high number of thermal cycles that the porous media is subject to. In the stationary technique, a single piece of foam can be used to stabilize a 4ame, but this is limited by 4ashback when ;ring at low rates. Adding a second piece of foam with a smaller pore size upstream of the ;rst piece acts as a stabilizing holder for the 4ame and prevents 4ashback [6]. The ;ne pores act as a 4ame arrestor by quenching the 4ame due to an increased di2usion of species and heat loss to the walls. Flame propagation is only possible when the rate of heat release from the reaction is higher than the heat loss to the surroundings. It has been found that the minimum
SL SMPS T UFL
laminar 4ame speed (cm/s) Scanning Mobility Particle Sizer temperature (K) upper 4ammability limit
Greek symbols
thermal di2usivity of gas mixture (m2 =s) e7ciency (%) equivalence ratio
pore size to prevent quenching inside a porous medium is only possible when the PLeclet number exceeds a certain limit [11,12]. By selecting a porous medium with a PLeclet number less than the critical number for the upstream region and a Pe greater than the critical number for the combustion region, the 4ame will be stabilized near the interface. Several studies have examined lean and ultra-lean combustion inside an inert porous media burner with two separate layers of ceramic foam [7,13–16]. In these studies an extension of the lean 4ammability limit and higher 4ame velocities were observed compared to laminar free 4ames. In certain applications, e.g. in the automobile industry or in domestic systems, size, simplicity and longevity are of primary concern. Steady combustion in porous media may o2er these. Rich 4ames in two-layer burners have not been examined before, to the authors’ knowledge, and it is interesting to see if the advantages of these devices can be used for partial oxidation applications. In the present work, stationary rich 4ames are stabilized in a compact two layer burner comprising di2erent porous materials. Methanol, iso-octane and automotive grade petrol, which are all potential fuels to be converted to H2 in the transport sector, are tested. Methane, which is the major constituent of natural gas, is also tested as it is likely to have applications in stationary systems. The speci;c objectives of this paper are to: (1) study the stability of 4ames inside several di2erent types of porous media to examine the possibility of extending the upper 4ammability limits, (2) test the robustness and operational lifetime of the porous media and (3) to examine the e2ectiveness of using this method to produce syngas. In the next Section, the apparatus and experimental techniques are described, while the results are presented and discussed in Section 3. The paper closes with a summary of the more important conclusions. 2. Experimental methods 2.1. Apparatus The burner was a quartz tube with an internal diameter of 30:4 mm, and an external diameter of 33:6 mm, as
H. Pedersen-Mjaanes et al. / International Journal of Hydrogen Energy 30 (2005) 579 – 592
581
approximately 13.7% MgO, 34.9% Al2 O3 and 51.4% SiO2 . Cordierite is known for its low heat expansion and high resistance to thermal shock. Fifteen percent of the earth’s crust is composed of alumina, Al2 O3 , making it a very abundant and potentially inexpensive material. Alumina has a higher thermal conductivity and melting point than cordierite but also a higher thermal expansivity and lower resistance to thermal shock which needs to be considered when selecting burner material. The properties of the porous alumina media used in the burners are tabulated in Table 1. 2.2. Measurement techniques
Fig. 1. Schematic of porous media burner.
illustrated in Fig. 1. The porous media was supported by a stainless steel ring attached to the top of the vaporizing nozzle (discussed below). Alumina paper was wrapped around the porous foam to prevent gases leaking around the edges and to absorb some of the thermal expansion of the porous media. A ;ne stainless steel mesh was placed at the exit of the vaporizing nozzle to act as a 4ame trap preventing 4ash back. The outside of the burner was insulated with 40 mm of Superwool607 and 50 mm of Rockwool insulation. Liquid fuel was pumped by a peristaltic pump to a commercial fuel vaporizer consisting of an injector housing and an electrically heated tube. The amount of heat delivered by the vaporizer was controlled by a variable transformer. Dry air was supplied by the laboratory’s compressor to the injector housing. The air 4ow rate was measured by a rotameter prior to entering the vaporizer. A diaphragm pressure gauge measured the pressure downstream of the rotameter so the rotameter reading could be corrected for changes in pressure. The vaporized fuel air mixture then entered the porous matrix burner where it was combusted. The fuel delivery system was calibrated with a “bucket and stopwatch” method to an estimated uncertainty of ±2%, while the rotameter was quoted as giving an uncertainty of ±2% of full scale de4ection. Tests were conducted using three di2erent types of porous media inside the burner: cordierite foam, alumina foam and alumina beads. Cordierite ceramics are composed of
Two thermocouples were used to measure temperature and determine if the system was stable: a K-type thermocouple (T1) imbedded inside the vaporizer to give the temperature of the fresh mixture entering the burner and an R-type thermocouple (T2) located 10 mm above the ceramic. The R-type thermocouple was corrected for radiation e2ects at high temperatures by balancing the radiative losses and gains with the convective heat transfer from the hot gases. The radiation heat 4ux form the burner was estimated, based on material emissivities from Ref. [7]. The products exiting the burner were sampled by a water cooled probe, passed through a ;lter to remove particulates, dried and injected into a gas chromatograph via a vacuum pump. A Perkin-Elmer AutoSystem XL Gas Chromatograph equipped with two columns (or stationary phases) and with a Flame Ionization Detector (FID) and a thermal conductivity detector (TCD) was used to measure the composition of the products. A porous polymer bead column was used to separate out the low molecular weight compounds (H2 ; CH4 ; CO; N2 ), prior to switching the valve which transfers the gas mixture to a molecular sieve column. This column was then used to separate out the larger molecules including CO2 and the various hydrocarbons. Based on results from equilibrium calculations (discussed later), the GC was calibrated for H2 , CO, CO2 , N2 , CH4 , C2 H2 , C2 H4 and C2 H6 . The porous polymer bead column was a 1=8 in diameter and 5 m long stainless steel column packed with HayeSepJ N. The second column was a 1=8 in diameter and 1 m long stainless steel column packed with molecular sieve 5A. Due to the large amount of unburnt hydrocarbons and soot in some of the experimental conditions, a scanning mobility particle sizer (SMPS) was used to measure the size and concentration of particles in the combustion products of octane. Using a modi;ed method for operating these instruments, the measurement time was on the order of 1 min [17]. 2.3. Experimental procedure and data analysis At the start of each experiment the vaporizer was heated to its operating point with just air passing through it. Once the vaporizer was above the vaporization temperature of the fuel, given in Table 2, the fuel pump was activated and the
582
H. Pedersen-Mjaanes et al. / International Journal of Hydrogen Energy 30 (2005) 579 – 592
Table 1 Properties of porous media used in the burners Properties
Foam burner
Length (mm) Porosity (%) Pores per cm Diameter (mm) a Data
Bead burner
Layer 1
Layer 2
Layer 1
Layer 2
23.6 86 24 —
47.2 84a 8 —
20 75 — 2–3
80 75b — 3–3.5
from manufacturer. Beads surface was covered with micropores.
b Measured.
Table 2 Temperature of reactants prior to injection into burner Thermocouple T1 (K) a Ambient
CH3 OH
CH4
C8 H18
353
293a
403
temperature, as vaporizer not required.
air–fuel mixture 4owed into the burner. The mixture was ignited with a butane torch at the burner’s exit. A mixture slightly richer than stoichiometric and a low mass 4ow rate was used so that the 4ame propagated into the porous medium. After the 4ame had stabilized at the interface between the support layer and the combustion layer, the 4ow was held constant for several minutes to allow the burner to warm up. After the warm up period, the 4ows were adjusted to their desired values for the experiment. After an additional period of several minutes the system reached thermal equilibrium, which can be deduced by constant thermocouple readings and a stationary 4ame front. At this point, a gas sample was injected into the GC and the products’ composition recorded by TotalChrom, the gas chromatograph’s software. This method of operation is to be contrasted to ;ltration combustion [8–10], where a propagating wave was induced. The measured results were compared to theoretical equilibrium calculations using NASA’s Chemical Equilibrium with Applications code (CEA) [18]. The code was run either with the adiabatic 4ame temperature corresponding to the of the experiment or at the measured products’ temperature T2. These results were converted to a dry basis to compare to the results obtained from the GC. 3. Results and discussion 3.1. Global behaviour and stability limits 3.1.1. Methanol Methanol 4ames would stabilize between the two differently sized pored ceramics. The UFL of methanol was
increased from 4.1 to 6.1 inside the foam burner and to 9.3 inside the bead burner. Based on a LHV of 21:6 × 106 J=kg for CH3 OH, a ;ring range of 135 to 2670 kW=m2 was attained at an equivalence ratio of 3.8. This corresponds to a turndown ratio of 20:1. This high turndown ratio is possible because of the wide range of gas mixture speeds that the 4ame can stabilize at (e.g. Fig. 2). Trends suggest that even greater turndown ratios would be possible closer to stoichiometric conditions due to the broadening of the operating envelope, as has been observed for lean 4ames by Hsu et al. [6] and also predicted by numerical models [7]. The maximum equivalence ratios and conversion e7ciencies for the various fuels tested are tabulated in Table 3. 3.1.2. Methane Methane 4ames stabilized at the interface up to a of 1.9, 20% higher than for a laminar free 4ame. It was determined that after the initial warm-up period a time of up to 10 min was required to ensure that the 4ame was stationary. Richer, but unstable, 4ames were also observed, where the combustion wave was seen to propagate downstream at a velocity on the order of 10−5 m=s, in agreement with Babkin et al. [11]. 3.1.3. Iso-octane and petrol Octane and petrol both behaved similarly to the other fuels tested and stabilized at the interface. However, compared to a methanol or methane 4ame which only takes a couple of minutes to enter the foam, the two hydrocarbon fuels took approximately 10 min to warm up the foam enough to propagate into it. This can be partially explained by iso-octane’s large quenching diameter, approximately 4 mm compared to 1:5 mm for methanol and 2:0 mm for methane [19]. However once stable, the upper 4ammability limit for the fuels could be attained. At equivalence ratios above 2.0 soot exiting the burner was visible, especially so for the petrol. The soot accumulated on the thermocouple and turned the ceramics from white to black. However, the soot could be removed by burning a mixture with excess oxygen for several minutes.
H. Pedersen-Mjaanes et al. / International Journal of Hydrogen Energy 30 (2005) 579 – 592
583
60
Flame Speed (cm/s)
50
40
30
20
10
0 3
4
5
6 7 Equivalence Ratio
8
9
10
Fig. 2. Operating envelope of methanol 4ame inside the bead burner. Table 3 Maximum equivalence ratios and conversion e7ciencies reached in the foam and bead burners
Methanol Methane Iso-Octane Petrol a Data
Free 4amea (UFL)
Alumina foam
Alumina beads
Conversion e7ciency (%)b (foam/beads)
4.08 1.64 3.8–4.2 —
6.1 1.9 4.0 —
9.3 — — 4.1
56/66 45/— 36/— 36/—
obtained from Refs. [23,24]. by Eq. (1).
b De;ned
3.2. Durability Cordierite foam was tested but it failed after several hours of operation. Alumina, with a higher melting point, did not melt or charr, but still su2ered serious damage. After about 20 h of testing the foam started to crumble. This can be explained by two mechanisms: (1) because its thermal coef;cient of expansion is approximately ten times higher than that of quartz, stresses built up in the foam during expansion and contraction, and (2) sudden thermal gradients within the material due to ignition, cooling down periods, and changes in equivalence ratios could cause weakening or fracturing within the material. This is more pronounced in the alumina as it has a lower thermal shock resistance than cordierite. The alumina beads appeared to su2er no degradation even after more than 100 h of testing. This is possibly due to
the lack of rigid links between the beads which resulted in lower thermal stresses caused by thermal cycles, making the bead burner considerably more robust. It can be concluded that, generally, beads will have a longer operational life than foams. However, the lifetime of the ceramic foams can be extended by selecting a burner material that has a similar thermal expansion rate and/or ceramics with a higher thermal shock resistance. 3.3. Product composition 3.3.1. Methanol Fig. 3 compares the adiabatic 4ame temperature of methanol, calculated by the equilibrium code CEA, with the temperature of the measured products. It is evident that the products are below the adiabatic temperature, especially at
584
H. Pedersen-Mjaanes et al. / International Journal of Hydrogen Energy 30 (2005) 579 – 592 2500 EQU. AD. T. FOAM T. BEAD T.
Temperature (K)
2000
1500
1000
500
0 1
2
3
4
5 6 7 Equivalence Ratio
8
9
10
Fig. 3. Calculated adiabatic methanol 4ame temperature compared to product temperature measured at burner exit. m˙ fuel was 4:6 × 10−2 g=s for all , except at = 4:0 for the foam burner and at = 6:5 for the bead burner where m˙ fuel was 3:6 × 10−2 g=s.
low . This is attributed to higher radiation heat losses from the burner at low . It is also probable that superadiabatic temperatures have been reached for both burners at a of 4.0, since at the point of measurement the products have already lost some heat to their surroundings and still have temperatures higher than the adiabatic 4ame temperature. Figs. 4 and 5 show that up to an equivalence ratio of 3.0, in the foam burner, the mole fraction of hydrogen and carbon monoxide increased with increasing equivalence ratio while the mole fraction of carbon dioxide decreased, in agreement with the equilibrium calculations. No hydrocarbons were measured in the products at equivalence ratios less than 2.5. Changing the height of the sampling probe above the ceramic, at a of 2.5, did not result in any signi;cant change in the product species or produce any discernable trends. This is a good indication that the products have reached equilibrium by the time they exit the ceramic and no more reactions are taking place. However, it appears that at higher equivalence ratios (e.g. ¿ 3) the reaction has moved away from equilibrium (e.g. Fig. 4 for H2 and Fig. 5 for CO and CO2 ). At a of 6.1, the largest mole fraction of hydrocarbons detected was methane at 0.6% with minute amounts of other hydrocarbons, not enough to account for the low (relative to the equilibrium) concentration of hydrogen measured. The product composition from the bead burner followed the equilibrium calculations and H2 increased with increasing up to a of 7.5. At higher the reaction moved away
from equilibrium as it did at the higher equivalence ratios in the foam. Fig. 5 also illustrates that CO and CO2 follow the equilibrium calculations relatively well until a of 6.0, when they start to deviate. In order to explore the deviation from equilibrium at very rich equivalence ratios, the CEA code was run with a set temperature equal to the measured one. The H2 concentration thus calculated is also plotted in Fig. 4, revealing that the measured composition lies between the two equilibrium calculations. At lower equivalence ratios, the high temperatures imply that equilibrium conditions are reached quicker, while the 4ames at very rich are probably in4uenced by ;nite residence time e2ects. The presence of small amounts of hydrocarbons observed in methanol 4ames is consistent with equilibrium expectations. As can be seen in Fig. 6, the equilibrium composition of methanol products at = 3:0 shows an increase in H2 O and CH4 mole fractions at temperature below 950 K with a corresponding decrease in CO and H2 . This change in product species can be attributed to the reverse steam reforming reaction CH4 + H2 O ↔ CO + 3H2 becoming important. The measured temperatures of the products for the range of operation of the foam burner was always above this critical temperature and minimum methane was measured. For ¿ 6:5 the bead burner products could fall below 950 K and small amounts of methane were indeed detected. Fig. 6 suggests that to maximize the H2 yield the product temperatures should not drop below 950 K.
H. Pedersen-Mjaanes et al. / International Journal of Hydrogen Energy 30 (2005) 579 – 592
585
0.45 H2 EQU. AD. T. H2 EQU. BEAD T. H2 FOAM H2 BEAD
0.4 0.35
Mole Fraction
0.3 0.25 0.2 0.15 0.1 0.05 0
1
2
3
4
5 6 Equivalence Ratio
7
8
9
10
Fig. 4. Equilibrium H2 species concentrations for methanol compared to experimental results. The black squares depict the H2 fraction calculated at the adiabatic 4ame temperature, while the black triangles depict the H2 fraction calculated at the temperature measured by the R-type thermocouple. m˙ fuel was 4:6 × 10−2 g=s for all , except at = 4:0 for the foam burner and at = 6:5 for the bead burner where m˙ fuel was 3:6 × 10−2 g=s.
0.25 CO EQU. AD. T. CO FOAM CO BEAD CO2 EQU. AD. T CO2 FOAM CO2 BEAD
Mole Fraction
0.2
0.15
0.1
0.05
0
1
2
3
4
5 6 Equivalence Ratio
7
8
9
10
Fig. 5. Equilibrium CO and CO2 species concentrations for methanol compared to experimental results. m˙ fuel was 4:6 × 10−2 g=s for all , except at = 4:0 for the foam burner and at = 6:5 for the bead burner where m˙ fuel was 3:6 × 10−2 g=s.
586
H. Pedersen-Mjaanes et al. / International Journal of Hydrogen Energy 30 (2005) 579 – 592 0.35
0.3
H2 EQU CO EQU CO2 EQU H2O EQU CH4 EQU
Mole Fraction
0.25
0.2
0.15
0.1
0.05
0 600
700
800
900
1000 1100 Temperature (K)
1200
1300
1400
1500
Fig. 6. Calculated equilibrium product species for methanol at various temperatures at = 3:0.
3.3.2. Methane The major combustion products compared to the predicted product composition from the equilibrium code is presented in Fig. 7. Overall, the experimental results follow equilibrium predictions for increasing H2 and CO with and decreasing CO2 with increasing albeit at a reduced rate. Changing the height of the sample probe did not produce any discernable trends in the products’ concentrations, indicating that the reaction had reached steady state. Methane and acetylene mole fractions increased from 0.4% at a of 1.55 to over 1.0% at a of 1.85. Increasing the mass 4ow rate resulted in higher temperatures that drove the reaction towards equilibrium producing less CH4 and C2 H2 which increased the H2 yield [20]. In an e2ort to understand the generation of hydrocarbons observed in the products, equilibrium calculations were performed at various temperatures. As illustrated in Fig. 8, CH4 is only generated at temperatures less than 1000 K. However, during these tests the products’ temperature remained considerably above this critical temperature. Hence, the signi;cant CH4 and C2 H2 species measured are due to ;nite rate kinetic e2ects, rather than the reverse steam reforming reaction. 3.3.3. Iso-octane and petrol The compositions of the products obtained with octane and petrol were equivalent within the accepted uncertainty of the measurements. Hydrogen production peaked at 10% for petrol and at a slightly higher 11% for octane, much lower than the equilibrium values. When the equivalence
ratio was increased beyond approximately 1.5 the actual hydrogen mole fraction started to diverge from the hydrogen predicted by the equilibrium code as illustrated in Fig. 9. Soot was visible at above 2.0 for petrol and above 3.0 for octane. Fig. 10 shows that in contrast to CH3 OH and to a greater extent than CH4 , signi;cant amounts of hydrocarbon species were present in the products. Methane, acetylene and ethylene were the most abundant hydrocarbons detected by the GC. Methane started to increase at a of 1.5 and continued to increase with increasing . Acetylene increased rapidly from a of 1.5, peaked at a of 2.0 and then slowly decreased thereafter. Ethylene started increasing at a of 2.5 and also continued to increase with . Equilibrium cannot explain the presence of these higher hydrocarbons. Fig. 11 shows that for a of 3.0, the steam reforming reaction becomes important for C8 H18 combustion at temperatures below 1100 K. However, during these tests the temperature at the point of extraction remained above this temperature. Therefore, since such high values cannot be generated by an equilibrium reaction, it is evident that the results are in4uenced by ;nite rate kinetics. Measurements at higher and lower 4ow rates are necessary to elucidate this point further. Performing a carbon balance with the measured species reveals that at higher equivalence ratios not all the carbon is accounted for. The additional carbon is tied up in higher hydrocarbons and soot not measured by the GC. To con;rm this, as well as to examine the nature of the soot
H. Pedersen-Mjaanes et al. / International Journal of Hydrogen Energy 30 (2005) 579 – 592
587
0.18
0.16
0.14
H2 EQU. AD. T. H2 EQU. FOAM T. H2 FOAM CO EQU. AD. T. CO FOAM CO2 EQU. AD. T. CO2 FOAM
Mole Fraction
0.12
0.1
0.08
0.06
0.04
0.02 1.5
1.55
1.6
1.65 1.7 1.75 Equivalence Ratio
1.8
1.85
1.9
Fig. 7. Equilibrium H2 , CO and CO2 species concentrations for methane compared to experimental results. EQU. AD. T. is the species concentration calculated with the equilibrium code at the adiabatic 4ame temperature. EQU. FOAM T. is the species concentration calculated with the equilibrium code at the temperature measured by the R-type thermocouple. m˙ fuel for = 1:55 to = 1:75 was 6:6 × 10−3 g=s and m˙ fuel for = 1:80 and = 1:85 was 5:5 × 10−3 g=s.
0.25
Mole Fraction
0.2
H2 EQU. CO EQU. CO2 EQU. H2O EQU. CH4 EQU.
0.15
0.1
0.05
0 600
700
800
900
1000 1100 Temperature (K)
1200
1300
1400
1500
Fig. 8. Calculated equilibrium product species for methane at various temperatures at = 2:0.
588
H. Pedersen-Mjaanes et al. / International Journal of Hydrogen Energy 30 (2005) 579 – 592 0.35 H2 EQU. AD. T. H2 EQU. FOAM T. H2 FOAM CO EQU. AD. T. CO FOAM CO2 EQU. AD. T CO2 FOAM
0.3
Mole Fraction
0.25
0.2
0.15
0.1
0.05
0
1
1.5
2
2.5 3 3.5 Equivalence Ratio
4
4.5
5
Fig. 9. Equilibrium H2 , CO and CO2 species concentrations for octane compared to experimental results. EQU. AD. T. is the species concentration calculated with the equilibrium code at the adiabatic 4ame temperature. EQU. FOAM T. is the species concentration calculated with the equilibrium code at the temperature measured by the R-type thermocouple. m˙ fuel for all was 1:5 × 10−2 g=s.
produced, a scanning mobility particle sizer was used to measure the particle number density and size in the products. It was observed that at stoichiometric conditions the amount of particles detected was negligible, as would be expected since all the carbon was being oxidized. Fig. 12 illustrates that the amount of soot increased with . The mass 4ow rate of reactants at a of 1.5 and 2.5 are similar, while the mass 4ow rate of reactants at a of 2.0 is approximately 70% higher. Although at a of 2.0 the particle concentration was higher than at a of 2.5, the total mass of particles was less. This test illustrates that by varying the residence time of the gases inside the combustion zone the particle concentration and particle size can be a2ected. Keeping constant and decreasing the mass 4ow rate, decreased the particle concentration but increased the particle size. Using petrol as fuel it was observed that as the distance separating the probe and ceramics changed, the products’ composition also changed [21]. When the probe height was increased the mole fraction of H2 , CO and minor hydrocarbon species all decreased, while the mole fraction of CO2 increased. This is because the system had not reached equilibrium and chemical reactions were still occurring. As the distance increased, the reaction moved closer to completion as more species became oxidized. This has future implications as for determining the ideal length of the burner.
3.4. Discussion It was observed that in porous media combustion, an operating envelope is possible as opposed to a unique burning velocity for each as occurs in a premixed laminar free 4ame (see Fig. 2), consistent with theoretical predictions [7]. Consider the various 4ame speeds at a of 3.8. The expected 4ame velocity will be higher than that for a free 4ame because of the aforementioned enhanced preheating of the reactants. If the 4ow of reactants is reduced the 4ame will attempt to move further upstream, but because of the ;ne pored ceramic will not be able to propagate into it. Additionally, the higher upstream heat losses will mean less energy is available to preheat the incoming gases. These two mechanisms will reduce the 4ame’s speed and allow it to stabilize at the interface as long as the amount of heat being released by the reaction balances the amount lost to the surroundings. If the reactants’ 4ow is increased the 4ame will propagate downstream away from the ;ne pored ceramic where more heat will be released, resulting in increased preheating of the reactants. The increased mass 4ow rate will also increase the amount of energy being released by the reactants. Both these mechanisms increase the 4ame’s velocity and act as a restoring force on the 4ame, moving it back towards the interface. We have observed experimentally the predictions by Diamantis et al. [7] that the self restoring force is only
H. Pedersen-Mjaanes et al. / International Journal of Hydrogen Energy 30 (2005) 579 – 592
589
0.09 CH4 EQU. AD. T. CH4 EQU. FOAM T. CH4 FOAM C2H2 EQU. AD. T. C2H2 FOAM C2H4 EQU. AD. T C2H4 FOAM
0.08 0.07
Mole Fraction
0.06 0.05 0.04 0.03 0.02 0.01 0 1
1.5
2
2.5
3 3.5 Equivalence Ratio
4
4.5
5
Fig. 10. Equilibrium CH4 , C2 H2 and C2 H4 species concentrations for octane compared to experimental results. EQU. AD. T. is the species concentration calculated with the equilibrium code at the adiabatic 4ame temperature. EQU. FOAM T. is the species concentration calculated with the equilibrium code at the temperature measured by the R-type thermocouple. m˙ fuel for all was 1:5 × 10−2 g=s.
0.35
0.3
H2 EQU CO EQU CO2 EQU H2O EQU CH4 EQU
Mole Fraction
0.25
0.2
0.15
0.1
0.05
0 600
700
800
900
1000 1100 1200 Temperature (K)
1300
1400
1500
Fig. 11. Calculated equilibrium product species for octane at various temperatures at = 3:0.
590
H. Pedersen-Mjaanes et al. / International Journal of Hydrogen Energy 30 (2005) 579 – 592 4
x 10
Equiv=1.5 Equiv=2.0 Equiv=2.0 Equiv=2.0 Equiv=2.5 Equiv=2.5 Equiv=2.5
3
Number Concentration (#/cm )
15
10
5
0 0
50
100
150
200 250 Particle Size (nm)
300
350
400
Fig. 12. Particle number concentration versus size for octane at various in a foam burner. EQUIV = 2:0 are three separate tests at the same operating conditions: = 2:0 and m˙ fuel = 2:0 × 10−2 g=s. Likewise EQUIV = 2:5 are three separate tests at the following conditions: = 2:5 and m˙ fuel = 1:5 × 10−2 g=s.
applicable in approximately the ;rst half of the porous matrix. If a positive disturbance is introduced in the downstream section of the matrix the 4ame will propagate to the exit and blowo2 will occur, with a slowly moving combustion wave. As expected, the envelope of stable operation narrows with increasing . As increases, the energy released by the reaction per unit mass of fuel decreases as less O2 is available to complete the process (see Fig. 3). This results in less energy being available to preheat the incoming gases, contributing to slower 4ame speeds, which in turn limits the range of stable operating points. In comparison to the bead burner, the foam burner’s operating envelope did not extend as far to the right and was narrower for the same . This is surprising because theoretically the foams should be able to outperform the beads due to their higher thermal conductivity and lower optical thickness (the greater the thickness the greater the attenuation of the radiation), resulting in overall superior heat transfer to the incoming gases [14]. Tests of di2erently sized pored foams and di2erently sized beads need to be conducted to determine why the beads outperformed the foams in these tests. The maximum amount of hydrogen produced from methanol was 42% by volume. The other major product apart from nitrogen was 22% CO which can be used in an additional water–gas shift reaction and converted to CO2 and H2 . Hydrogen production increases with increasing
as the reaction moves further away from completion. As increases, the amount of nitrogen is decreasing which accounts for part of the higher molar fraction of hydrogen in the products. Therefore, the burner may have a higher e7ciency of converting the energy inside the fuel to energy stored in H2 at a lower than the maximum achievable. De;ning the function of the burner as that of converting a fuel to H2 , and assuming that the CO produced can help convert the H2 O to H2 in a future water–gas shift reaction, then the e7ciency of the burner can be given by Eq. (1): =
m˙ H2 LHVH2 + m˙ CO LHVCO : m˙ fuel LHVfuel
(1)
A similar de;nition was used in Refs. [22] and [2], where the LHV of H2 produced was divided by the LHV of fuel to calculate conversion e7ciency (which implies that the water–gas shift reaction has taken place). The highest conversion e7ciency of methanol to syngas, based on Eq. (1), was 56% at a of 4.5 in the foam burner and 66% at a of 3.8 in the bead burner. The maximum syngas conversion e7ciency for methane was 45%. Petrol and iso-octane achieved conversion e7ciencies of 36% at a of 1.5. However, when the additional hydrogen stored in CH4 , C2 H2 , C2 H4 and C2 H6 is taken into account the potential hydrogen e7ciency peaks at a much higher 75% at a = 4:0. The considerable amount of H2 trapped in H2 O
H. Pedersen-Mjaanes et al. / International Journal of Hydrogen Energy 30 (2005) 579 – 592
and the hydrocarbons needs to be recovered to make this a viable technology for extracting H2 from petrol. Although both the present technique of stable 4ames and the ;ltration combustion technique can increase the upper 4ammability limits, the latter technique has been able to achieve higher at least with respect to methane. Comparing the results speci;cally to those of Bingue et al. [9], similar amounts of H2 , CO and CO2 are produced for the same range of tested. Since there are currently no results published for methanol or octane for ;ltration combustion, a direct comparison cannot be made. In the present experiments for octane, we have achieved equivalence ratios higher than the theoretical optimal for H2 production (3.125; found by simply balancing the reaction of fuel and oxygen to give only CO and H2 in the products). The higher 4ammability limits achievable with ;ltration combustion may therefore not be as important. The simplicity and smaller size of the present technique may favour it for automobile applications. 4. Conclusions Steady rich 4ames of methanol, methane, iso-octane and petrol were stabilized at the interface between two di2erently sized pored ceramic foams over a range of equivalence ratios. Methanol 4ames were also stabilized over a range of equivalence ratios inside a burner using alumina beads as the porous media. The upper 4ammability limits of methanol and methane were increased by combustion inside the porous media. At a minimum, the conventional rich 4ammability limits for petrol and iso-octane were also reached. A large operating envelope was observed making high turndown ratios possible. The maximum mole fraction of hydrogen was 28% from methanol, 13% from methane and 11% from octane inside the porous foam burner. This corresponds to a conversion e7ciency of 56%, 45% and 36% respectively when comparing the LHV of H2 and CO in the products to the LHV of the fuel. Considerable amounts of CH4 , C2 H2 and C2 H4 were measured in the combustion products of methane, octane and petrol at higher equivalence ratios which cannot be explained by equilibrium conditions, hence suggesting ;nite rate e2ects. The maximum mole fraction of hydrogen was 42% from methanol inside the alumina bead burner which is equivalent to a conversion e7ciency of 66%. These results show that steady rich combustion inside a porous medium can be used to reform liquid and gaseous fuels into syngas. Quick startup times, compact size and high turn down ratios make it a potential candidate for the next generation fuel cell powered automobile. Acknowledgements The authors wish to acknowledge the support of the Chevening Scholarship Foundation, Bombardier and the Cambridge University Engineering Department. We
591
also thank Dr. G. Biskes for assistance with the soot measurements.
References [1] Meyer A, Schroll C, Lesieur R. Development and evaluation of multi-fuel cell power plant for transportation applications. In Fuel Cell Technology for Vehicles, volume PT-84 of SAE International Progress in Technology Series, 2000. p. 111–9. [2] Ahmed S, Krumpelt M. Hydrogen from hydrocarbon fuels for fuel cells. Int J Hydrogen Energy 2001;26:291–301. [3] Ersoz A, Olgun H, Ozdogan S, Gungor C, Akgun F, Tlrls M. Autothermal reforming as a hydrocarbon fuel processing option for pem fuel cell. J Power Sources 2003;118(1–2): 384–92. [4] Weinberg FJ, Bartleet TG, Carleton FB, Rimbotti P. Partial oxidation of fuel-rich mixtures in a spouted bed combustor. Combust Flame 1988;72:235–9. [5] Egerton A, Gugan K, Weinberg J. The mechanism of smouldering in cigarettes. Combust Flame 1963;7:63–78. [6] Hsu P, Evans W, Howell J. Experimental and numerical study of premixed combustion within nonhomogenous porous ceramics. Combust Sci Tech 1993;90:149–72. [7] Diamantis DJ, Mastorakos E, Goussis DA. Simulations of premixed combustion in porous media. Combust Theory Model 2002;6:383–411. [8] Drayton MK, Saveliev AV, Kennedy LA, Fridman AA, Li Y. Syngas production using superadiabatic combustion of ultra-rich methane-air mixtures. Proc. Comb. Inst. 1998;27:1361–7. [9] Bingue JP, Saveliev AV, Fridman AA, Kennedy LA. Hydrogen production in ultra-rich ;ltration combustion of methane and hydrogen sul;de. Int J Hydrogen Energy 2002;27:643–9. [10] Zhdanok SA. Porous media combustion based hydrogen production. In European Combustion Meeting, Orleans, 2003. [11] Babkin VS, Korzhavin AA, Bunev VA. Propagation of premixed gaseous explosion 4ames in porous media. Combust Flame 1991;87:182–90. [12] Trimis D, Wawrzinek K. Flame stabilization of highly di2usive gas mixtures in porous inert media. In The 12th International Conference on Fluid Flow Technologies Budapest, Hungary (CMFF’03), 2003. p. 61–8. [13] Trimis D, Durst F. Combustion in a porous medium-advances and applications. Combust Sci Technol 1996;121:153–68. [14] Trimis D, Durst F, Pickencker O, Pickencker K. Porous medium combustor versus combustion systems with free 4ames. In Second International Symposium on Heat Transfer Enhancement and Energy Conservation, 1997. p. 339–45. [15] Zepter K, Hustad JE. Combustion and emission studies in a porous inert media burner. Meeting of the Scandinavian-Nordic Section of the Combustion Institute, 2001. [16] Mathis Jr. WM, Ellzey JL. Flame stabilization, operating range, and emissions for a methane/air porous burner. Combust Sci Technol 2003;175:825–39. [17] Biskos G. Theoretical and experimental investigation of the di2erential mobility spectrometer. Ph.D. thesis, University of Cambridge, 2004.
592
H. Pedersen-Mjaanes et al. / International Journal of Hydrogen Energy 30 (2005) 579 – 592
[18] Gordon S, McBride BJ. Computer program for calculation of complex chemical equilibrium compositions and applications. Technical Report Part I: Analysis, NASA RP-1311, 1994. [19] Glassman I. Combustion, 3rd ed. San Diego: Academic Press; 1996. [20] Chan L. Hydrogen production by rich 4ame combustion for fuel cells. 4th year technical report, Engineering Department, University of Cambridge, 2003. [21] Pedersen-Mjaanes H. Hydrogen production from rich combustion. Master’s thesis, Engineering Department, University of Cambridge, 2003.
[22] Hagan M, Northrop W, Bowers B, Rumsey J, Prabhu S. Automotive fuel processing for PEM fuel cells. In Fuel Cell Technology for Vehicles, volume PT-84 of SAE International Progress in Technology Series, 2000. p. 103–10. [23] Turns SR. An introduction to combustion, 2nd ed. New York: McGraw-Hill; 2000. [24] API. Alcohols: A technical assessment of their application as motor fuels. Technical Report No. 4261, American Petroleum Institute, 1976.
www.elsevier.com/locate/ijhydene
Hydrogen production from rich combustion in porous media H. Pedersen-Mjaanes, L. Chan, E. Mastorakos∗ Department of Engineering, University of Cambridge, Cambridge, CB2 1PZ, UK Accepted 5 May 2004 Available online 15 July 2004
Abstract This paper examines rich combustion of methanol, methane, octane and automotive-grade petrol inside inert porous media in an e2ort to examine the suitability of the concept for hydrogen production. Species concentrations were measured and operating limits were tested of steady rich 4ames stabilized inside a two-layer alumina foam burner and a two-layer alumina bead burner. Using a conversion e7ciency based on lower heating values, up to 56% of the methanol was converted to syngas (H2 , CO) inside the alumina foam burner and 66% inside the alumina bead burner. Using the same e7ciency de;nition, 45% percent of the methane and 36% of the octane and petrol was converted to syngas with a signi;cant portion of the energy remaining trapped in CH4 , C2 H2 and C2 H4 . For methanol, the highest equivalence ratio observed for stable combustion was 6.1 inside the foam burner and 9.3 inside the bead burner which are higher than the conventional upper 4ammability limit (UFL) of 4.1. Methane’s UFL was increased to 1.9 and, at a minimum, the conventional upper 4ammability limits of iso-octane and petrol were attained. A wide operating envelope was observed, which allowed for large turndown ratios up to 20:1. The composition of the products of the methanol 4ames examined here were close to equilibrium for relatively low equivalence ratios, while those of hydrocarbon 4ames di2ered signi;cantly from equilibrium for all suggesting that ;nite rate kinetics are important. The high conversion e7ciencies, quick startup times, compact size, and the absence of a catalyst make the present burner suitable for consideration as part of a reformer in a fuel cell powered automobile. ? 2004 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved. Keywords: Porous media burner; Hydrogen; Rich combustion; Partial oxidation
1. Introduction Hydrogen contains more energy per unit mass than any other fuel and produces minimum emissions when combusted and essentially no emissions when electrochemically converted to electricity in a fuel cell. It has the potential to reduce dependence on imported fuels depending on how it is generated. These are a few of the reasons why hydrogen is considered the fuel of the future. However, making the transition to a hydrogen economy means converting the existing fossil fuel infrastructure that is already developed. An alternative is to use a fuel reformer to extract hydrogen from existing fuels and/or alcohols. For example, ∗
Corresponding author. Tel.: +44-1223-332690; fax: +441223-332065. E-mail addresses: [email protected] (H. Pedersen-Mjaanes), [email protected] (E. Mastorakos).
an automobile could be powered by any number of fuels currently available, including methanol, ethanol, gasoline and even diesel; they could operate more e7ciently and produce fewer emissions, at the same time creating a bridge between the future hydrogen economy and the present fossil fuel economy. Although numerous technologies exist for producing hydrogen, the three main methods are steam reforming, autothermal reforming and partial oxidation (rich combustion). Partial oxidation and autothermal reforming processes have better dynamic responses and are more compact than a comparative steam reformer [1–3]. Usually, they involve a catalyst which ensures good conversion, but a catalyst is prone to poisoning and requires great care with the sulphur content of the fuel and with particulates. One of the most promising non-catalytic partial oxidation techniques is combustion inside a porous medium. This medium could be a 4uidized or a stationary bed. In a
0360-3199/$ 30.00 ? 2004 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.ijhydene.2004.05.006
580
H. Pedersen-Mjaanes et al. / International Journal of Hydrogen Energy 30 (2005) 579 – 592
Nomenclature A CEA D dm k L LHV m˙ P Pe PPC
cross sectional area (m2 ) Chemical Equilibrium with Applications diameter (m) characteristic pore size (m) thermal conductivity (W=m K) length (m) lower heating value (kJ/kg) mass 4ow rate (kg/s) pressure (kPa) PLeclet number pores per centimetre
4uidized bed system, the incoming velocity of the gases is high enough to lift the particles. A spouted-bed combustor involves motion of solid inert (or catalytic) particles in a recirculatory manner, this technique has been used to stabilise rich 4ames of methane in order to produce syngas [4]. Combustion in stationary porous media occurs inside the voids of the porous matrix. Energy is transferred to the solid by convection from the hot gases. Conduction and radiation through the solid porous medium, which has much higher thermal conductivity than the gas mixture and additionally radiates signi;cantly, distributes some of this energy to the region upstream of the 4ame. This in turn transfers energy via convection to the incoming reactants. The porous matrix acts in essence as an integral preheater. This preheating allows for higher combustion temperatures and increased stability compared to a free premixed laminar 4ame. Locally, the 4ame zone may reach temperatures higher than the adiabatic 4ame temperature of the unburnt mixture, which gives the name “superadiabatic” to this combustion technology [5–10]. The same e2ect is also produced by the recirculation of solid particles in the spouted-bed combustor of Weinberg et al. [4]. Filtration combustion inside a stationary, inert porous medium has also been used to produce syngas [8–10]. In this technique, a con;ned wave was allowed to propagate along the solid, controlled by periodically reversing of the reactants 4ow. Currently, several gaseous fuels and kerosene have been tested using this method. A disadvantage of this technique compared to a stationary 4ame system may be its relatively large size and high number of thermal cycles that the porous media is subject to. In the stationary technique, a single piece of foam can be used to stabilize a 4ame, but this is limited by 4ashback when ;ring at low rates. Adding a second piece of foam with a smaller pore size upstream of the ;rst piece acts as a stabilizing holder for the 4ame and prevents 4ashback [6]. The ;ne pores act as a 4ame arrestor by quenching the 4ame due to an increased di2usion of species and heat loss to the walls. Flame propagation is only possible when the rate of heat release from the reaction is higher than the heat loss to the surroundings. It has been found that the minimum
SL SMPS T UFL
laminar 4ame speed (cm/s) Scanning Mobility Particle Sizer temperature (K) upper 4ammability limit
Greek symbols
thermal di2usivity of gas mixture (m2 =s) e7ciency (%) equivalence ratio
pore size to prevent quenching inside a porous medium is only possible when the PLeclet number exceeds a certain limit [11,12]. By selecting a porous medium with a PLeclet number less than the critical number for the upstream region and a Pe greater than the critical number for the combustion region, the 4ame will be stabilized near the interface. Several studies have examined lean and ultra-lean combustion inside an inert porous media burner with two separate layers of ceramic foam [7,13–16]. In these studies an extension of the lean 4ammability limit and higher 4ame velocities were observed compared to laminar free 4ames. In certain applications, e.g. in the automobile industry or in domestic systems, size, simplicity and longevity are of primary concern. Steady combustion in porous media may o2er these. Rich 4ames in two-layer burners have not been examined before, to the authors’ knowledge, and it is interesting to see if the advantages of these devices can be used for partial oxidation applications. In the present work, stationary rich 4ames are stabilized in a compact two layer burner comprising di2erent porous materials. Methanol, iso-octane and automotive grade petrol, which are all potential fuels to be converted to H2 in the transport sector, are tested. Methane, which is the major constituent of natural gas, is also tested as it is likely to have applications in stationary systems. The speci;c objectives of this paper are to: (1) study the stability of 4ames inside several di2erent types of porous media to examine the possibility of extending the upper 4ammability limits, (2) test the robustness and operational lifetime of the porous media and (3) to examine the e2ectiveness of using this method to produce syngas. In the next Section, the apparatus and experimental techniques are described, while the results are presented and discussed in Section 3. The paper closes with a summary of the more important conclusions. 2. Experimental methods 2.1. Apparatus The burner was a quartz tube with an internal diameter of 30:4 mm, and an external diameter of 33:6 mm, as
H. Pedersen-Mjaanes et al. / International Journal of Hydrogen Energy 30 (2005) 579 – 592
581
approximately 13.7% MgO, 34.9% Al2 O3 and 51.4% SiO2 . Cordierite is known for its low heat expansion and high resistance to thermal shock. Fifteen percent of the earth’s crust is composed of alumina, Al2 O3 , making it a very abundant and potentially inexpensive material. Alumina has a higher thermal conductivity and melting point than cordierite but also a higher thermal expansivity and lower resistance to thermal shock which needs to be considered when selecting burner material. The properties of the porous alumina media used in the burners are tabulated in Table 1. 2.2. Measurement techniques
Fig. 1. Schematic of porous media burner.
illustrated in Fig. 1. The porous media was supported by a stainless steel ring attached to the top of the vaporizing nozzle (discussed below). Alumina paper was wrapped around the porous foam to prevent gases leaking around the edges and to absorb some of the thermal expansion of the porous media. A ;ne stainless steel mesh was placed at the exit of the vaporizing nozzle to act as a 4ame trap preventing 4ash back. The outside of the burner was insulated with 40 mm of Superwool607 and 50 mm of Rockwool insulation. Liquid fuel was pumped by a peristaltic pump to a commercial fuel vaporizer consisting of an injector housing and an electrically heated tube. The amount of heat delivered by the vaporizer was controlled by a variable transformer. Dry air was supplied by the laboratory’s compressor to the injector housing. The air 4ow rate was measured by a rotameter prior to entering the vaporizer. A diaphragm pressure gauge measured the pressure downstream of the rotameter so the rotameter reading could be corrected for changes in pressure. The vaporized fuel air mixture then entered the porous matrix burner where it was combusted. The fuel delivery system was calibrated with a “bucket and stopwatch” method to an estimated uncertainty of ±2%, while the rotameter was quoted as giving an uncertainty of ±2% of full scale de4ection. Tests were conducted using three di2erent types of porous media inside the burner: cordierite foam, alumina foam and alumina beads. Cordierite ceramics are composed of
Two thermocouples were used to measure temperature and determine if the system was stable: a K-type thermocouple (T1) imbedded inside the vaporizer to give the temperature of the fresh mixture entering the burner and an R-type thermocouple (T2) located 10 mm above the ceramic. The R-type thermocouple was corrected for radiation e2ects at high temperatures by balancing the radiative losses and gains with the convective heat transfer from the hot gases. The radiation heat 4ux form the burner was estimated, based on material emissivities from Ref. [7]. The products exiting the burner were sampled by a water cooled probe, passed through a ;lter to remove particulates, dried and injected into a gas chromatograph via a vacuum pump. A Perkin-Elmer AutoSystem XL Gas Chromatograph equipped with two columns (or stationary phases) and with a Flame Ionization Detector (FID) and a thermal conductivity detector (TCD) was used to measure the composition of the products. A porous polymer bead column was used to separate out the low molecular weight compounds (H2 ; CH4 ; CO; N2 ), prior to switching the valve which transfers the gas mixture to a molecular sieve column. This column was then used to separate out the larger molecules including CO2 and the various hydrocarbons. Based on results from equilibrium calculations (discussed later), the GC was calibrated for H2 , CO, CO2 , N2 , CH4 , C2 H2 , C2 H4 and C2 H6 . The porous polymer bead column was a 1=8 in diameter and 5 m long stainless steel column packed with HayeSepJ N. The second column was a 1=8 in diameter and 1 m long stainless steel column packed with molecular sieve 5A. Due to the large amount of unburnt hydrocarbons and soot in some of the experimental conditions, a scanning mobility particle sizer (SMPS) was used to measure the size and concentration of particles in the combustion products of octane. Using a modi;ed method for operating these instruments, the measurement time was on the order of 1 min [17]. 2.3. Experimental procedure and data analysis At the start of each experiment the vaporizer was heated to its operating point with just air passing through it. Once the vaporizer was above the vaporization temperature of the fuel, given in Table 2, the fuel pump was activated and the
582
H. Pedersen-Mjaanes et al. / International Journal of Hydrogen Energy 30 (2005) 579 – 592
Table 1 Properties of porous media used in the burners Properties
Foam burner
Length (mm) Porosity (%) Pores per cm Diameter (mm) a Data
Bead burner
Layer 1
Layer 2
Layer 1
Layer 2
23.6 86 24 —
47.2 84a 8 —
20 75 — 2–3
80 75b — 3–3.5
from manufacturer. Beads surface was covered with micropores.
b Measured.
Table 2 Temperature of reactants prior to injection into burner Thermocouple T1 (K) a Ambient
CH3 OH
CH4
C8 H18
353
293a
403
temperature, as vaporizer not required.
air–fuel mixture 4owed into the burner. The mixture was ignited with a butane torch at the burner’s exit. A mixture slightly richer than stoichiometric and a low mass 4ow rate was used so that the 4ame propagated into the porous medium. After the 4ame had stabilized at the interface between the support layer and the combustion layer, the 4ow was held constant for several minutes to allow the burner to warm up. After the warm up period, the 4ows were adjusted to their desired values for the experiment. After an additional period of several minutes the system reached thermal equilibrium, which can be deduced by constant thermocouple readings and a stationary 4ame front. At this point, a gas sample was injected into the GC and the products’ composition recorded by TotalChrom, the gas chromatograph’s software. This method of operation is to be contrasted to ;ltration combustion [8–10], where a propagating wave was induced. The measured results were compared to theoretical equilibrium calculations using NASA’s Chemical Equilibrium with Applications code (CEA) [18]. The code was run either with the adiabatic 4ame temperature corresponding to the of the experiment or at the measured products’ temperature T2. These results were converted to a dry basis to compare to the results obtained from the GC. 3. Results and discussion 3.1. Global behaviour and stability limits 3.1.1. Methanol Methanol 4ames would stabilize between the two differently sized pored ceramics. The UFL of methanol was
increased from 4.1 to 6.1 inside the foam burner and to 9.3 inside the bead burner. Based on a LHV of 21:6 × 106 J=kg for CH3 OH, a ;ring range of 135 to 2670 kW=m2 was attained at an equivalence ratio of 3.8. This corresponds to a turndown ratio of 20:1. This high turndown ratio is possible because of the wide range of gas mixture speeds that the 4ame can stabilize at (e.g. Fig. 2). Trends suggest that even greater turndown ratios would be possible closer to stoichiometric conditions due to the broadening of the operating envelope, as has been observed for lean 4ames by Hsu et al. [6] and also predicted by numerical models [7]. The maximum equivalence ratios and conversion e7ciencies for the various fuels tested are tabulated in Table 3. 3.1.2. Methane Methane 4ames stabilized at the interface up to a of 1.9, 20% higher than for a laminar free 4ame. It was determined that after the initial warm-up period a time of up to 10 min was required to ensure that the 4ame was stationary. Richer, but unstable, 4ames were also observed, where the combustion wave was seen to propagate downstream at a velocity on the order of 10−5 m=s, in agreement with Babkin et al. [11]. 3.1.3. Iso-octane and petrol Octane and petrol both behaved similarly to the other fuels tested and stabilized at the interface. However, compared to a methanol or methane 4ame which only takes a couple of minutes to enter the foam, the two hydrocarbon fuels took approximately 10 min to warm up the foam enough to propagate into it. This can be partially explained by iso-octane’s large quenching diameter, approximately 4 mm compared to 1:5 mm for methanol and 2:0 mm for methane [19]. However once stable, the upper 4ammability limit for the fuels could be attained. At equivalence ratios above 2.0 soot exiting the burner was visible, especially so for the petrol. The soot accumulated on the thermocouple and turned the ceramics from white to black. However, the soot could be removed by burning a mixture with excess oxygen for several minutes.
H. Pedersen-Mjaanes et al. / International Journal of Hydrogen Energy 30 (2005) 579 – 592
583
60
Flame Speed (cm/s)
50
40
30
20
10
0 3
4
5
6 7 Equivalence Ratio
8
9
10
Fig. 2. Operating envelope of methanol 4ame inside the bead burner. Table 3 Maximum equivalence ratios and conversion e7ciencies reached in the foam and bead burners
Methanol Methane Iso-Octane Petrol a Data
Free 4amea (UFL)
Alumina foam
Alumina beads
Conversion e7ciency (%)b (foam/beads)
4.08 1.64 3.8–4.2 —
6.1 1.9 4.0 —
9.3 — — 4.1
56/66 45/— 36/— 36/—
obtained from Refs. [23,24]. by Eq. (1).
b De;ned
3.2. Durability Cordierite foam was tested but it failed after several hours of operation. Alumina, with a higher melting point, did not melt or charr, but still su2ered serious damage. After about 20 h of testing the foam started to crumble. This can be explained by two mechanisms: (1) because its thermal coef;cient of expansion is approximately ten times higher than that of quartz, stresses built up in the foam during expansion and contraction, and (2) sudden thermal gradients within the material due to ignition, cooling down periods, and changes in equivalence ratios could cause weakening or fracturing within the material. This is more pronounced in the alumina as it has a lower thermal shock resistance than cordierite. The alumina beads appeared to su2er no degradation even after more than 100 h of testing. This is possibly due to
the lack of rigid links between the beads which resulted in lower thermal stresses caused by thermal cycles, making the bead burner considerably more robust. It can be concluded that, generally, beads will have a longer operational life than foams. However, the lifetime of the ceramic foams can be extended by selecting a burner material that has a similar thermal expansion rate and/or ceramics with a higher thermal shock resistance. 3.3. Product composition 3.3.1. Methanol Fig. 3 compares the adiabatic 4ame temperature of methanol, calculated by the equilibrium code CEA, with the temperature of the measured products. It is evident that the products are below the adiabatic temperature, especially at
584
H. Pedersen-Mjaanes et al. / International Journal of Hydrogen Energy 30 (2005) 579 – 592 2500 EQU. AD. T. FOAM T. BEAD T.
Temperature (K)
2000
1500
1000
500
0 1
2
3
4
5 6 7 Equivalence Ratio
8
9
10
Fig. 3. Calculated adiabatic methanol 4ame temperature compared to product temperature measured at burner exit. m˙ fuel was 4:6 × 10−2 g=s for all , except at = 4:0 for the foam burner and at = 6:5 for the bead burner where m˙ fuel was 3:6 × 10−2 g=s.
low . This is attributed to higher radiation heat losses from the burner at low . It is also probable that superadiabatic temperatures have been reached for both burners at a of 4.0, since at the point of measurement the products have already lost some heat to their surroundings and still have temperatures higher than the adiabatic 4ame temperature. Figs. 4 and 5 show that up to an equivalence ratio of 3.0, in the foam burner, the mole fraction of hydrogen and carbon monoxide increased with increasing equivalence ratio while the mole fraction of carbon dioxide decreased, in agreement with the equilibrium calculations. No hydrocarbons were measured in the products at equivalence ratios less than 2.5. Changing the height of the sampling probe above the ceramic, at a of 2.5, did not result in any signi;cant change in the product species or produce any discernable trends. This is a good indication that the products have reached equilibrium by the time they exit the ceramic and no more reactions are taking place. However, it appears that at higher equivalence ratios (e.g. ¿ 3) the reaction has moved away from equilibrium (e.g. Fig. 4 for H2 and Fig. 5 for CO and CO2 ). At a of 6.1, the largest mole fraction of hydrocarbons detected was methane at 0.6% with minute amounts of other hydrocarbons, not enough to account for the low (relative to the equilibrium) concentration of hydrogen measured. The product composition from the bead burner followed the equilibrium calculations and H2 increased with increasing up to a of 7.5. At higher the reaction moved away
from equilibrium as it did at the higher equivalence ratios in the foam. Fig. 5 also illustrates that CO and CO2 follow the equilibrium calculations relatively well until a of 6.0, when they start to deviate. In order to explore the deviation from equilibrium at very rich equivalence ratios, the CEA code was run with a set temperature equal to the measured one. The H2 concentration thus calculated is also plotted in Fig. 4, revealing that the measured composition lies between the two equilibrium calculations. At lower equivalence ratios, the high temperatures imply that equilibrium conditions are reached quicker, while the 4ames at very rich are probably in4uenced by ;nite residence time e2ects. The presence of small amounts of hydrocarbons observed in methanol 4ames is consistent with equilibrium expectations. As can be seen in Fig. 6, the equilibrium composition of methanol products at = 3:0 shows an increase in H2 O and CH4 mole fractions at temperature below 950 K with a corresponding decrease in CO and H2 . This change in product species can be attributed to the reverse steam reforming reaction CH4 + H2 O ↔ CO + 3H2 becoming important. The measured temperatures of the products for the range of operation of the foam burner was always above this critical temperature and minimum methane was measured. For ¿ 6:5 the bead burner products could fall below 950 K and small amounts of methane were indeed detected. Fig. 6 suggests that to maximize the H2 yield the product temperatures should not drop below 950 K.
H. Pedersen-Mjaanes et al. / International Journal of Hydrogen Energy 30 (2005) 579 – 592
585
0.45 H2 EQU. AD. T. H2 EQU. BEAD T. H2 FOAM H2 BEAD
0.4 0.35
Mole Fraction
0.3 0.25 0.2 0.15 0.1 0.05 0
1
2
3
4
5 6 Equivalence Ratio
7
8
9
10
Fig. 4. Equilibrium H2 species concentrations for methanol compared to experimental results. The black squares depict the H2 fraction calculated at the adiabatic 4ame temperature, while the black triangles depict the H2 fraction calculated at the temperature measured by the R-type thermocouple. m˙ fuel was 4:6 × 10−2 g=s for all , except at = 4:0 for the foam burner and at = 6:5 for the bead burner where m˙ fuel was 3:6 × 10−2 g=s.
0.25 CO EQU. AD. T. CO FOAM CO BEAD CO2 EQU. AD. T CO2 FOAM CO2 BEAD
Mole Fraction
0.2
0.15
0.1
0.05
0
1
2
3
4
5 6 Equivalence Ratio
7
8
9
10
Fig. 5. Equilibrium CO and CO2 species concentrations for methanol compared to experimental results. m˙ fuel was 4:6 × 10−2 g=s for all , except at = 4:0 for the foam burner and at = 6:5 for the bead burner where m˙ fuel was 3:6 × 10−2 g=s.
586
H. Pedersen-Mjaanes et al. / International Journal of Hydrogen Energy 30 (2005) 579 – 592 0.35
0.3
H2 EQU CO EQU CO2 EQU H2O EQU CH4 EQU
Mole Fraction
0.25
0.2
0.15
0.1
0.05
0 600
700
800
900
1000 1100 Temperature (K)
1200
1300
1400
1500
Fig. 6. Calculated equilibrium product species for methanol at various temperatures at = 3:0.
3.3.2. Methane The major combustion products compared to the predicted product composition from the equilibrium code is presented in Fig. 7. Overall, the experimental results follow equilibrium predictions for increasing H2 and CO with and decreasing CO2 with increasing albeit at a reduced rate. Changing the height of the sample probe did not produce any discernable trends in the products’ concentrations, indicating that the reaction had reached steady state. Methane and acetylene mole fractions increased from 0.4% at a of 1.55 to over 1.0% at a of 1.85. Increasing the mass 4ow rate resulted in higher temperatures that drove the reaction towards equilibrium producing less CH4 and C2 H2 which increased the H2 yield [20]. In an e2ort to understand the generation of hydrocarbons observed in the products, equilibrium calculations were performed at various temperatures. As illustrated in Fig. 8, CH4 is only generated at temperatures less than 1000 K. However, during these tests the products’ temperature remained considerably above this critical temperature. Hence, the signi;cant CH4 and C2 H2 species measured are due to ;nite rate kinetic e2ects, rather than the reverse steam reforming reaction. 3.3.3. Iso-octane and petrol The compositions of the products obtained with octane and petrol were equivalent within the accepted uncertainty of the measurements. Hydrogen production peaked at 10% for petrol and at a slightly higher 11% for octane, much lower than the equilibrium values. When the equivalence
ratio was increased beyond approximately 1.5 the actual hydrogen mole fraction started to diverge from the hydrogen predicted by the equilibrium code as illustrated in Fig. 9. Soot was visible at above 2.0 for petrol and above 3.0 for octane. Fig. 10 shows that in contrast to CH3 OH and to a greater extent than CH4 , signi;cant amounts of hydrocarbon species were present in the products. Methane, acetylene and ethylene were the most abundant hydrocarbons detected by the GC. Methane started to increase at a of 1.5 and continued to increase with increasing . Acetylene increased rapidly from a of 1.5, peaked at a of 2.0 and then slowly decreased thereafter. Ethylene started increasing at a of 2.5 and also continued to increase with . Equilibrium cannot explain the presence of these higher hydrocarbons. Fig. 11 shows that for a of 3.0, the steam reforming reaction becomes important for C8 H18 combustion at temperatures below 1100 K. However, during these tests the temperature at the point of extraction remained above this temperature. Therefore, since such high values cannot be generated by an equilibrium reaction, it is evident that the results are in4uenced by ;nite rate kinetics. Measurements at higher and lower 4ow rates are necessary to elucidate this point further. Performing a carbon balance with the measured species reveals that at higher equivalence ratios not all the carbon is accounted for. The additional carbon is tied up in higher hydrocarbons and soot not measured by the GC. To con;rm this, as well as to examine the nature of the soot
H. Pedersen-Mjaanes et al. / International Journal of Hydrogen Energy 30 (2005) 579 – 592
587
0.18
0.16
0.14
H2 EQU. AD. T. H2 EQU. FOAM T. H2 FOAM CO EQU. AD. T. CO FOAM CO2 EQU. AD. T. CO2 FOAM
Mole Fraction
0.12
0.1
0.08
0.06
0.04
0.02 1.5
1.55
1.6
1.65 1.7 1.75 Equivalence Ratio
1.8
1.85
1.9
Fig. 7. Equilibrium H2 , CO and CO2 species concentrations for methane compared to experimental results. EQU. AD. T. is the species concentration calculated with the equilibrium code at the adiabatic 4ame temperature. EQU. FOAM T. is the species concentration calculated with the equilibrium code at the temperature measured by the R-type thermocouple. m˙ fuel for = 1:55 to = 1:75 was 6:6 × 10−3 g=s and m˙ fuel for = 1:80 and = 1:85 was 5:5 × 10−3 g=s.
0.25
Mole Fraction
0.2
H2 EQU. CO EQU. CO2 EQU. H2O EQU. CH4 EQU.
0.15
0.1
0.05
0 600
700
800
900
1000 1100 Temperature (K)
1200
1300
1400
1500
Fig. 8. Calculated equilibrium product species for methane at various temperatures at = 2:0.
588
H. Pedersen-Mjaanes et al. / International Journal of Hydrogen Energy 30 (2005) 579 – 592 0.35 H2 EQU. AD. T. H2 EQU. FOAM T. H2 FOAM CO EQU. AD. T. CO FOAM CO2 EQU. AD. T CO2 FOAM
0.3
Mole Fraction
0.25
0.2
0.15
0.1
0.05
0
1
1.5
2
2.5 3 3.5 Equivalence Ratio
4
4.5
5
Fig. 9. Equilibrium H2 , CO and CO2 species concentrations for octane compared to experimental results. EQU. AD. T. is the species concentration calculated with the equilibrium code at the adiabatic 4ame temperature. EQU. FOAM T. is the species concentration calculated with the equilibrium code at the temperature measured by the R-type thermocouple. m˙ fuel for all was 1:5 × 10−2 g=s.
produced, a scanning mobility particle sizer was used to measure the particle number density and size in the products. It was observed that at stoichiometric conditions the amount of particles detected was negligible, as would be expected since all the carbon was being oxidized. Fig. 12 illustrates that the amount of soot increased with . The mass 4ow rate of reactants at a of 1.5 and 2.5 are similar, while the mass 4ow rate of reactants at a of 2.0 is approximately 70% higher. Although at a of 2.0 the particle concentration was higher than at a of 2.5, the total mass of particles was less. This test illustrates that by varying the residence time of the gases inside the combustion zone the particle concentration and particle size can be a2ected. Keeping constant and decreasing the mass 4ow rate, decreased the particle concentration but increased the particle size. Using petrol as fuel it was observed that as the distance separating the probe and ceramics changed, the products’ composition also changed [21]. When the probe height was increased the mole fraction of H2 , CO and minor hydrocarbon species all decreased, while the mole fraction of CO2 increased. This is because the system had not reached equilibrium and chemical reactions were still occurring. As the distance increased, the reaction moved closer to completion as more species became oxidized. This has future implications as for determining the ideal length of the burner.
3.4. Discussion It was observed that in porous media combustion, an operating envelope is possible as opposed to a unique burning velocity for each as occurs in a premixed laminar free 4ame (see Fig. 2), consistent with theoretical predictions [7]. Consider the various 4ame speeds at a of 3.8. The expected 4ame velocity will be higher than that for a free 4ame because of the aforementioned enhanced preheating of the reactants. If the 4ow of reactants is reduced the 4ame will attempt to move further upstream, but because of the ;ne pored ceramic will not be able to propagate into it. Additionally, the higher upstream heat losses will mean less energy is available to preheat the incoming gases. These two mechanisms will reduce the 4ame’s speed and allow it to stabilize at the interface as long as the amount of heat being released by the reaction balances the amount lost to the surroundings. If the reactants’ 4ow is increased the 4ame will propagate downstream away from the ;ne pored ceramic where more heat will be released, resulting in increased preheating of the reactants. The increased mass 4ow rate will also increase the amount of energy being released by the reactants. Both these mechanisms increase the 4ame’s velocity and act as a restoring force on the 4ame, moving it back towards the interface. We have observed experimentally the predictions by Diamantis et al. [7] that the self restoring force is only
H. Pedersen-Mjaanes et al. / International Journal of Hydrogen Energy 30 (2005) 579 – 592
589
0.09 CH4 EQU. AD. T. CH4 EQU. FOAM T. CH4 FOAM C2H2 EQU. AD. T. C2H2 FOAM C2H4 EQU. AD. T C2H4 FOAM
0.08 0.07
Mole Fraction
0.06 0.05 0.04 0.03 0.02 0.01 0 1
1.5
2
2.5
3 3.5 Equivalence Ratio
4
4.5
5
Fig. 10. Equilibrium CH4 , C2 H2 and C2 H4 species concentrations for octane compared to experimental results. EQU. AD. T. is the species concentration calculated with the equilibrium code at the adiabatic 4ame temperature. EQU. FOAM T. is the species concentration calculated with the equilibrium code at the temperature measured by the R-type thermocouple. m˙ fuel for all was 1:5 × 10−2 g=s.
0.35
0.3
H2 EQU CO EQU CO2 EQU H2O EQU CH4 EQU
Mole Fraction
0.25
0.2
0.15
0.1
0.05
0 600
700
800
900
1000 1100 1200 Temperature (K)
1300
1400
1500
Fig. 11. Calculated equilibrium product species for octane at various temperatures at = 3:0.
590
H. Pedersen-Mjaanes et al. / International Journal of Hydrogen Energy 30 (2005) 579 – 592 4
x 10
Equiv=1.5 Equiv=2.0 Equiv=2.0 Equiv=2.0 Equiv=2.5 Equiv=2.5 Equiv=2.5
3
Number Concentration (#/cm )
15
10
5
0 0
50
100
150
200 250 Particle Size (nm)
300
350
400
Fig. 12. Particle number concentration versus size for octane at various in a foam burner. EQUIV = 2:0 are three separate tests at the same operating conditions: = 2:0 and m˙ fuel = 2:0 × 10−2 g=s. Likewise EQUIV = 2:5 are three separate tests at the following conditions: = 2:5 and m˙ fuel = 1:5 × 10−2 g=s.
applicable in approximately the ;rst half of the porous matrix. If a positive disturbance is introduced in the downstream section of the matrix the 4ame will propagate to the exit and blowo2 will occur, with a slowly moving combustion wave. As expected, the envelope of stable operation narrows with increasing . As increases, the energy released by the reaction per unit mass of fuel decreases as less O2 is available to complete the process (see Fig. 3). This results in less energy being available to preheat the incoming gases, contributing to slower 4ame speeds, which in turn limits the range of stable operating points. In comparison to the bead burner, the foam burner’s operating envelope did not extend as far to the right and was narrower for the same . This is surprising because theoretically the foams should be able to outperform the beads due to their higher thermal conductivity and lower optical thickness (the greater the thickness the greater the attenuation of the radiation), resulting in overall superior heat transfer to the incoming gases [14]. Tests of di2erently sized pored foams and di2erently sized beads need to be conducted to determine why the beads outperformed the foams in these tests. The maximum amount of hydrogen produced from methanol was 42% by volume. The other major product apart from nitrogen was 22% CO which can be used in an additional water–gas shift reaction and converted to CO2 and H2 . Hydrogen production increases with increasing
as the reaction moves further away from completion. As increases, the amount of nitrogen is decreasing which accounts for part of the higher molar fraction of hydrogen in the products. Therefore, the burner may have a higher e7ciency of converting the energy inside the fuel to energy stored in H2 at a lower than the maximum achievable. De;ning the function of the burner as that of converting a fuel to H2 , and assuming that the CO produced can help convert the H2 O to H2 in a future water–gas shift reaction, then the e7ciency of the burner can be given by Eq. (1): =
m˙ H2 LHVH2 + m˙ CO LHVCO : m˙ fuel LHVfuel
(1)
A similar de;nition was used in Refs. [22] and [2], where the LHV of H2 produced was divided by the LHV of fuel to calculate conversion e7ciency (which implies that the water–gas shift reaction has taken place). The highest conversion e7ciency of methanol to syngas, based on Eq. (1), was 56% at a of 4.5 in the foam burner and 66% at a of 3.8 in the bead burner. The maximum syngas conversion e7ciency for methane was 45%. Petrol and iso-octane achieved conversion e7ciencies of 36% at a of 1.5. However, when the additional hydrogen stored in CH4 , C2 H2 , C2 H4 and C2 H6 is taken into account the potential hydrogen e7ciency peaks at a much higher 75% at a = 4:0. The considerable amount of H2 trapped in H2 O
H. Pedersen-Mjaanes et al. / International Journal of Hydrogen Energy 30 (2005) 579 – 592
and the hydrocarbons needs to be recovered to make this a viable technology for extracting H2 from petrol. Although both the present technique of stable 4ames and the ;ltration combustion technique can increase the upper 4ammability limits, the latter technique has been able to achieve higher at least with respect to methane. Comparing the results speci;cally to those of Bingue et al. [9], similar amounts of H2 , CO and CO2 are produced for the same range of tested. Since there are currently no results published for methanol or octane for ;ltration combustion, a direct comparison cannot be made. In the present experiments for octane, we have achieved equivalence ratios higher than the theoretical optimal for H2 production (3.125; found by simply balancing the reaction of fuel and oxygen to give only CO and H2 in the products). The higher 4ammability limits achievable with ;ltration combustion may therefore not be as important. The simplicity and smaller size of the present technique may favour it for automobile applications. 4. Conclusions Steady rich 4ames of methanol, methane, iso-octane and petrol were stabilized at the interface between two di2erently sized pored ceramic foams over a range of equivalence ratios. Methanol 4ames were also stabilized over a range of equivalence ratios inside a burner using alumina beads as the porous media. The upper 4ammability limits of methanol and methane were increased by combustion inside the porous media. At a minimum, the conventional rich 4ammability limits for petrol and iso-octane were also reached. A large operating envelope was observed making high turndown ratios possible. The maximum mole fraction of hydrogen was 28% from methanol, 13% from methane and 11% from octane inside the porous foam burner. This corresponds to a conversion e7ciency of 56%, 45% and 36% respectively when comparing the LHV of H2 and CO in the products to the LHV of the fuel. Considerable amounts of CH4 , C2 H2 and C2 H4 were measured in the combustion products of methane, octane and petrol at higher equivalence ratios which cannot be explained by equilibrium conditions, hence suggesting ;nite rate e2ects. The maximum mole fraction of hydrogen was 42% from methanol inside the alumina bead burner which is equivalent to a conversion e7ciency of 66%. These results show that steady rich combustion inside a porous medium can be used to reform liquid and gaseous fuels into syngas. Quick startup times, compact size and high turn down ratios make it a potential candidate for the next generation fuel cell powered automobile. Acknowledgements The authors wish to acknowledge the support of the Chevening Scholarship Foundation, Bombardier and the Cambridge University Engineering Department. We
591
also thank Dr. G. Biskes for assistance with the soot measurements.
References [1] Meyer A, Schroll C, Lesieur R. Development and evaluation of multi-fuel cell power plant for transportation applications. In Fuel Cell Technology for Vehicles, volume PT-84 of SAE International Progress in Technology Series, 2000. p. 111–9. [2] Ahmed S, Krumpelt M. Hydrogen from hydrocarbon fuels for fuel cells. Int J Hydrogen Energy 2001;26:291–301. [3] Ersoz A, Olgun H, Ozdogan S, Gungor C, Akgun F, Tlrls M. Autothermal reforming as a hydrocarbon fuel processing option for pem fuel cell. J Power Sources 2003;118(1–2): 384–92. [4] Weinberg FJ, Bartleet TG, Carleton FB, Rimbotti P. Partial oxidation of fuel-rich mixtures in a spouted bed combustor. Combust Flame 1988;72:235–9. [5] Egerton A, Gugan K, Weinberg J. The mechanism of smouldering in cigarettes. Combust Flame 1963;7:63–78. [6] Hsu P, Evans W, Howell J. Experimental and numerical study of premixed combustion within nonhomogenous porous ceramics. Combust Sci Tech 1993;90:149–72. [7] Diamantis DJ, Mastorakos E, Goussis DA. Simulations of premixed combustion in porous media. Combust Theory Model 2002;6:383–411. [8] Drayton MK, Saveliev AV, Kennedy LA, Fridman AA, Li Y. Syngas production using superadiabatic combustion of ultra-rich methane-air mixtures. Proc. Comb. Inst. 1998;27:1361–7. [9] Bingue JP, Saveliev AV, Fridman AA, Kennedy LA. Hydrogen production in ultra-rich ;ltration combustion of methane and hydrogen sul;de. Int J Hydrogen Energy 2002;27:643–9. [10] Zhdanok SA. Porous media combustion based hydrogen production. In European Combustion Meeting, Orleans, 2003. [11] Babkin VS, Korzhavin AA, Bunev VA. Propagation of premixed gaseous explosion 4ames in porous media. Combust Flame 1991;87:182–90. [12] Trimis D, Wawrzinek K. Flame stabilization of highly di2usive gas mixtures in porous inert media. In The 12th International Conference on Fluid Flow Technologies Budapest, Hungary (CMFF’03), 2003. p. 61–8. [13] Trimis D, Durst F. Combustion in a porous medium-advances and applications. Combust Sci Technol 1996;121:153–68. [14] Trimis D, Durst F, Pickencker O, Pickencker K. Porous medium combustor versus combustion systems with free 4ames. In Second International Symposium on Heat Transfer Enhancement and Energy Conservation, 1997. p. 339–45. [15] Zepter K, Hustad JE. Combustion and emission studies in a porous inert media burner. Meeting of the Scandinavian-Nordic Section of the Combustion Institute, 2001. [16] Mathis Jr. WM, Ellzey JL. Flame stabilization, operating range, and emissions for a methane/air porous burner. Combust Sci Technol 2003;175:825–39. [17] Biskos G. Theoretical and experimental investigation of the di2erential mobility spectrometer. Ph.D. thesis, University of Cambridge, 2004.
592
H. Pedersen-Mjaanes et al. / International Journal of Hydrogen Energy 30 (2005) 579 – 592
[18] Gordon S, McBride BJ. Computer program for calculation of complex chemical equilibrium compositions and applications. Technical Report Part I: Analysis, NASA RP-1311, 1994. [19] Glassman I. Combustion, 3rd ed. San Diego: Academic Press; 1996. [20] Chan L. Hydrogen production by rich 4ame combustion for fuel cells. 4th year technical report, Engineering Department, University of Cambridge, 2003. [21] Pedersen-Mjaanes H. Hydrogen production from rich combustion. Master’s thesis, Engineering Department, University of Cambridge, 2003.
[22] Hagan M, Northrop W, Bowers B, Rumsey J, Prabhu S. Automotive fuel processing for PEM fuel cells. In Fuel Cell Technology for Vehicles, volume PT-84 of SAE International Progress in Technology Series, 2000. p. 103–10. [23] Turns SR. An introduction to combustion, 2nd ed. New York: McGraw-Hill; 2000. [24] API. Alcohols: A technical assessment of their application as motor fuels. Technical Report No. 4261, American Petroleum Institute, 1976.