The Influence of the Distributed Reaction Regime on Fuel Reforming Conditions

The distributed reaction regime explored previously has been shown to alter the distribution of reformate products, yielding more favorable results [1–4]. From these studies, a theory was formulated on how the distributed reaction regime influenced the reformate product composition and is reported within this paper.

Accurate kinetic modeling of reforming is still a topic of continuous development since it cannot capture the full range of reactions [5–8]. Most detailed kinetic combustion mechanisms are capable of predicting partial oxidation conditions, but do not have the capability to accurately represent noncatalytic steam reforming and soot formation reactions [8]. This limits the mechanism's accuracy to predict the effects of the distributed reaction regime. Jess [7] has shown significant interaction of the steam with soot and lighter (C₁–C₂) alkane molecules. However, a comprehensive model that includes both dry and steam reforming reactions for heavy alkane fuels has not been developed. Previous examination [5,9] by our group has successfully modeled only partial oxidation under conventional reforming conditions. These efforts have partially attempted to model distributed reforming condition using detailed combustion mechanism in a one-dimensional plug flow reactor with recirculation [2].

However, as the distributed reaction regime uses turbulent flow to control the chemistry through the entrainment of exhaust products, three-dimensional computational fluid dynamics (CFD) simulations can potentially offer a more accurate approach to modeling the distributed reaction regime than one-dimensional plug flow reactor with recirculation [10]. CFD is possible for simpler fuels such as methane or carbon [10–12]. However, detailed mechanisms [13–15] of a heavy hydrocarbon fuel capable of capturing reforming behavior were computationally too large for conventional CFD approaches. A reduced mechanism could be developed on steam and dry reforming kinetics provided sufficient experimental data were made available. In this paper, the activities of the reactions were determined through further analysis of the rather limited experimental data.

In noncatalytic reforming, five dominant reactions occur: partial oxidation, steam reforming, dry reforming, pyrolysis, and the water gas shift reactions. As kinetic mechanisms for JP8 or a suitable surrogate have only been developed for partial oxidation and pyrolysis conditions, experimental data from literature were assessed to determine the conditions that would support these reactions. Table 1 summarizes the temperatures and timescales, which activate these reactions. Sections 2.1–2.5 discuss this information in further detail. Significant literature is available on pyrolysis and partial oxidation reactions. However, limited research has been conducted in noncatalytic steam and dry reforming of hydrocarbons fuels such as JP8. Most information on noncatalytic steam and dry reforming reactions were derived from the experimental results of blank reactors used in catalyst evaluation and the gasification of biomass and wastes. Steam and dry reforming reactions were considered the most desirable, as they contributed the highest yields of syngas, and converted nonusable (H₂O and CO₂) gas phase species to favorable syngas (H₂ and CO). Partial oxidation was the second most desirable reaction, as it converted the fuel to a useable syngas, but generated lower yields of syngas than either steam or dry reforming reactions. In addition, oxidative reactions have the potential to oxidize the syngas reducing syngas yields. Reactions associated with thermal cracking/pyrolysis were the least desirable, as they generated hydrocarbons with stronger bonds (C₂H₂, C₂H₄) and also provided lowest syngas yields. Note that the above hydrocarbon reactions are also associated with the formation of soot that resulted in reduced syngas yields. The presence of soot not only decreases the syngas yields, efficiency, and reactor performance, but also is an environmental issue.

Partial oxidation is the exothermic reaction of a hydrocarbon fuel with reduced amounts of oxygen so that the reactions do not precede complete oxidation. This is a rapid reaction, occurring within timescales of 20–100 ms. Inadvertently, some hydrogen will oxidize to form steam, reducing syngas yields. Partial oxidation reformers are designed such that their residence times greatly exceed the timescales needed for partial oxidation reaction (20–100 ms), allowing steam reforming and water gas shift reactions to occur.

Al-Hamamre et al. [16] simulated the partial oxidation of methane with Chemkin code and the verified with the experimental data at O/C ratios of 1.68–1.92. Using injection temperatures of 400 °C, partial oxidation reactions occurred within 10 ms. Drayton et al. [17] evaluated the partial oxidation of methane at reactor temperatures of 1100–1400 °C, and O/C ratios of 0.5–2.0. Numerical simulations were used to support this work. Partial oxidation reactions occurred within 20 ms and accounted for 60% of the hydrogen generated. The remaining hydrogen was concluded to be from steam reforming and the water gas shift reactions. Vourliotakis et al. [18] presented data on the partial oxidation of ethanol at temperatures of 791–804 °C and O/C ratio of 1.5. Their results showed active partial oxidation reactions at timescales of 80 ms. Roth and Wirtz [19] evaluated the partial oxidation of diesel fuel using a reactor with a fixed residence time of 400 ms. The activity of partial oxidation reaction was affected by oxygen concentrations. Zhao et al. [20], Eisazadeh-Far et al. [21], and Tang et al. [22] reported that lowering the oxygen concentrations and dilution within the reactor reduced the activity of the partial oxidation reactions, which is a critical aspect of achieving the distributed reaction regime.

Steam reforming is the endothermic decomposition of a hydrocarbon fuel and steam to produce a hydrogen-rich syngas. It was observed that reactor temperatures of 800–1000 °C were required to activate steam reforming reactions [23–30] under noncatalytic conditions. Steam reforming reactions are slower, and generally require residence times greater than 400–500 ms. Results that are more favorable appeared when the reactor residence time was on the order of 1000 ms and temperatures between 1000 and 1300 °C. Studies [31,32] using residence times of 50–200 ms, more typical of catalytic reformers, indicated that steam reforming reactions were inactive at these temperatures and timescales. Catalysts enhanced the activity of the steam reforming reactions, allowing full conversion within shorter timescales of 50–200 ms. This results in the steam reforming reactions often being cited as inactive without the presence of catalysts [31,32].

Woodruf [33] demonstrated the steam gasification of char at temperatures of 1000–1050 °C, and residence time of 1000 ms. Molintas [34] showed steam reforming of tar could occur at temperatures of 800–900 °C. Noticeable steam reforming reactions were observed within 5–10 ms. Bartekova investigated the steam cracking (at steam to carbon ratio, S/C = 4.5) of hexadecane at timescales of 50–300 ms and reactor temperatures of 700–760 °C. Under those conditions, the formation of carbon monoxide and carbon dioxide was observed, which was indicative of steam reforming reactions. Jangsawang et al. [35] evaluated cellulose under high temperature air/steam gasification steam gasification conditions. Peak hydrogen concentrations occurred at temperatures of 900 °C.

Jess [7] evaluated toluene, naphthalene, and benzene under steam reforming conditions at temperatures of 700–1400 °C and residence times between 300 and 2000 ms. Steam reforming reactions, as indicated by the evolution of carbon monoxide and carbon dioxide, began to occur at temperatures greater than 1100 °C.

Steam reforming reactions appeared more active in the presence of oxygen and occurred at lower temperatures. Marty and Grouset [36] theorized that oxygen helped to break up stable hydrocarbon molecules, generating free radicals. These radicals then promoted the further decomposition of hydrocarbons and fostered steam reforming reactions. Kinetic studies of steam reforming conducted by Hiblot et al. [8] and Jess [7] supported Marty's [36] theory, as hydrogen was primarily generated through hydroxyl radicals (from steam), interacting primarily with the C₁–C₂ hydrocarbons. In addition, the heat generated by partial oxidation reactions will foster additional endothermic steam reforming reactions. Sharma and Schoegl [26] evaluated propane under partial oxidation, autothermal, and steam reforming conditions at reactor temperatures of 800–1000 °C and residence times of 1300 ms. The reactor was operated at O/C ratio of 0.42 under partial oxidation and autothermal conditions. Their data did not provide information on the S/C ratios examined. Slight increases in carbon monoxide and hydrogen concentrations were observed when steam was added under partial oxidation conditions, indicating steam reforming reactions were active. However, when operated under pure steam reforming mode (without oxygen), little activity in steam reforming reaction was overserved, as almost no carbon dioxide or carbon monoxide was generated. Products were primarily derived from thermal cracking.

Marty and Grouset [36] also evaluated propane under autothermal conditions, but at higher O/C ratio of 1.2 and S/C ratios of 1.5. Residence times were on order of 1000 ms, but yielded significantly higher amounts of hydrogen concentrations approaching 25%. Parmar [24] evaluated diesel in a blank reactor at O/C ratios of 0.4–1.0 and an S/C ratio of 1.5. The reactor was operated at temperatures of 700–850 °C, and residence time of 2830 ms. From the data (at 850 °C and O/C = 0.4), 25% more oxygen was detected in the molar flow rate of carbon monoxide and carbon dioxide than was available in the air. This additional oxygen was indicative of steam reforming reactions extracting oxygen from the steam releasing additional hydrogen. Roth and Wirtz [19] operated the reactor at residence times of 400 ms and a reactor temperature of 1300 °C. An increase in the steam to carbon ratio promoted an increase in carbon monoxide and hydrogen concentrations to reveal the presence of steam reforming reactions. The work of Scenna and Gupta [4] showed active steam reforming reactions with JP8 that occurred at timescales of 750–850 ms. Figure 1 shows that the addition of steam (S/C = 0–0.10) reduced the reactor temperatures from 1000 °C to 755 °C, while increasing the conversion from 90% to 97%, thus indicating activity in the endothermic steam reforming reactions. Further increase in steam content did not significantly improve the conversion or show such a rapid decrease in reactor temperature. Conversion is defined as the ratio of the molar flow rate of carbon oxidized to the molar flow rate of carbon content of the fuel, see the below equation:

Dry reforming is the interaction of carbon dioxide with a hydrocarbon fuel. It is considered a slow reaction. Dry reforming reactions are up to three times slower than steam reforming [37]. Data available in the literature often evaluated samples over a period of 15–20 min, or until each sample were completely converted [38]. However, the initial reactions occurred over a relatively short period of time. Dry reforming literature was limited, focusing primarily on waste and biomass feedstocks. In the following, various aspects of dry reforming are presented.

Dry reforming reactions required reactor temperatures of 800–1000 °C and timescales of at least 1000 ms to activate. Zhang et al. [39] observed meaningful conversion (10–80%) of methane through dry reforming reactions at temperatures of 1000–1200 °C. Residence time was on order of 2000 ms. The dry reformation of char from various biomass sources has been examined by a number of authors [27,40,41]. Reactor temperatures of 800–1000 °C were required to activate the dry reforming reactions. Barkia et al. [42] investigated the dry reforming of shale oil using thermal gravimetric analysis. Noticeable conversion occurred at reactor temperatures of 900 °C. No information was shown regarding reformate composition over time.

Thermal cracking is the endothermic thermal decomposition of hydrocarbons into smaller hydrocarbons without interaction with oxidizers. These reactions are significantly more active than steam or dry reforming reactions. For certain fuels, temperatures as low as 600–700 °C can activate cracking reactions in time duration as short as 50 ms. Thermal cracking reactions are also associated with deposit and soot formation [43].

A wider range of data was available for pyrolysis reactions. Bartekova and Bajus [44] investigated the thermal steam cracking of hexadecane. Steam was mixed with hexadecane at S/C ratio of 3. The reactants were evaluated at residence times of 5 to 300 ms and reactor temperatures of 700–780 °C. Noticeable conversion occurred at timescales of 5 ms and temperatures of 760 °C. Steam reforming was also observed in this reported work. The activity of the thermal cracking reactions was identified from the formation of lower hydrocarbons in the syngas (CH₄, C₂H₂, and C₂H₄) without any presence of oxygenated species (carbon monoxide, carbon dioxide). Longer residence times improved conversion and yields. Moler et al. [43] evaluated JP8 under pyrolysis conditions. Temperatures were varied between 350 and 1000 °C, over a period of 300 to 3000 ms. Ten percent conversion was observed at temperatures of 636 °C, over a time duration of 1000 ms. Higher temperatures resulted in rapid conversion. Temperature below 600 °C showed little activity. Moler et al. [43] observed that temperature had a strong impact on the produced gas composition distribution. Reactor temperatures of 600–700 °C favored low molecular weight simple hydrocarbons (C₁–C₄). Higher temperatures of 700–850 °C favored alkylated cycloalkanes. Even higher temperatures of 850–1000 °C promoted the formation of alkylated benzenes, naphthalenes, and cyclo-aromatics. Temperatures lower than 600 °C did not appear to activate the cracking reactions. Moler also evaluated the effect of stability additive in JP8 on reforming. The JFA-5 additive improved the stability at temperatures less than 350 °C, but showed no influence at temperatures over 600–1000 °C.

Zeppieri et al. [45] presented data on decane pyrolysis in conjunction with the modeling efforts. Experimental data were presented for residence times of up to 220 ms and temperatures from 646 to 757 °C. Activity in the pyrolysis reactions was observed as early as 40 ms. Kang et al. [46] evaluated hexane at longer residence times (1000–5000 ms) as compared to other experimental work. Hexane appeared to be more stable than hexadecane, JP8, or decane, requiring higher temperatures for conversion to occur. Activity in the pyrolysis reactions was overserved only at temperatures of 875 °C and timescales of 1000 ms.

The water gas shift is a mildly exothermic reaction, in which carbon monoxide reacts with steam to produce carbon dioxide and hydrogen. Limited research has been conducted on the homogeneous forward water gas shift under conditions relevant to this work. Graven and Long [47] evaluated the water gas shift reactions at 900 °C and residence times of ∼530–540 ms in a quartz reactor. They observed noticeable activity under these examined conditions. To isolate the forward water gas shift reaction from the reverse water gas shift reaction, they purposefully limited the extent of reaction to ∼0.5–2% by controlling the residence time. These authors also observed that the conversion increased linearly with time, up to 11%. This shows that some reactions will occur, influencing hydrogen concentrations by 1–2%. Bustamante-Londono [48] evaluated the forward and reverse water gas shift reaction at residence times of 300–500 ms, temperatures of 527–927 °C, and pressures of 100–1600 kpa using a quartz and Inconel reactor. While Inconel is typically not considered for its catalytic activity, some catalytic activity was observed with Inconel over that of quartz reactor. At pressures of 100 kpa, the Inconel reactor at temperatures of 527 °C yielded a conversion of about 8%, but at higher temperatures of ∼700 °C much higher conversion of about 70% was achieved. However, the quartz reactor only demonstrated 1% conversion at temperatures of 877–927 °C, with no activity observed at lower temperatures.

Conventional reforming occurs in three phases: (1) chemical decomposition, (2) oxidation, and (3) steam reformation. Figure 2 shows all three phases for an N-dodecane fuel (C₁₂H₂₆) at unity molar O/C ratio. Initially, the hydrocarbon fuel decomposed into multiple simpler hydrocarbons (CH₄, C₂H₂, and C₂H₄). This decomposition occurred rapidly near the front of the reactor, but can also occur to a lesser extent in subsequent phases. This was followed by a highly exothermic oxidative region, wherein smaller hydrocarbons generated in the first phase rapidly reacted with the available oxygen. This presented as a rapid increase in reactor temperatures, and the emergence of hydrogen, steam, and other combustion products. After the oxygen was consumed, the steam generated as a byproduct of the oxidative phase promoted the endothermic steam reforming of the remaining unconverted hydrocarbons. Those reactions were slower and occurred near the rear of the reactor. This last phase was denoted by decreasing concentrations of steam and reactor temperatures.

In the distributed reaction regime, the characteristic chemical time and length scales exceed characteristic time and length scales associated with turbulent transport. To initiate this, the fuel–air mixture was injected through a high velocity jet into the reactor. This jet entrained hot reactive products into the fuel–air mixture, which diluted the local oxygen concentrations. The dilution helped to reduce the activity of the oxidative reactions [22,49], which lengthened the chemical length and timescales. As partial oxidation undergoes a relatively rapid chemical reaction, a small reduction in activity is not expected to affect the overall conversion. The high velocity jet also enhanced mixing, which reduced the characteristic time and length scales associated with turbulent transport. More distributed conditions promoted greater entrainment of the hot reactive products into the fuel–air mixture, which altered the chemistry when ignition did occur. Without the enhanced mixing and dilution, a conventional flame would have emerged, as reactions would have proceeded at conventional time and length scales. Figure 3 shows the dilution of a premixed charge as a function of entrainment. As entrainment increased, the local fuel and oxygen concentrations diminished, while the local concentrations of hydrogen, steam, carbon monoxide, and carbon dioxide increased substantially.

The entrainment of hot reactive products into the reaction zone influenced the temperature distribution within the reactor. The entrainment raised the average reactor temperature, while the entrained products reduced peak temperatures through thermal dilution. Elevating the average reactor temperature enhanced the activity of the reforming reactions, while reducing the peak reactor temperature. This limited the thermal cracking reactions and reduced the thermal stresses on the reactor. A higher average temperature also served to stabilize the reactions under low or oxygen-depleted conditions.

The benefits from the distributed reaction regime were derived from the entrainment of the hot reactive product gases into the fuel–air mixture. In reforming, soot was primarily formed through hydrogen abstraction carbon addition (HACA) mechanism [50] (reactions R1–R2 given below), while acetylene was formed through dehydrogenation reactions [8] (reactions R3–R4 given below). In particular, steam and carbon dioxide were shown in combustion literature to suppress acetylene and soot formation [19,51–53]. Soot abatement was induced through dilution and chemical interactions of the entrained carbon dioxide and steam. Steam and carbon dioxide promoted the formation of hydroxyl radical, limiting the availability of hydrogen radicals, as shown in reactions R5–R7. The lack of hydrogen radicals interfered with acetylene formation and the HACA mechanism [52,54]. Conditions occurring within the flamelet in eddies regime caused the partial oxidation reactions to propagate faster, thus limiting entrainment. The more distributed conditions are believed to limit the activity of dehydrogenation reaction, which should cause the more distributed conditions to favor reformate products hydrocarbon with a higher H/C ratio.

Higher reactor temperatures of 800–1000 °C caused the entrained hot reactive products (steam and carbon dioxide) to promote steam and dry reforming reactions, enhancing reformate yield. The addition of steam to the reactants (wet partial oxidation) will increase this effect. Since more distributed conditions promoted greater entrainment, this offers an increased potential for steam and dry reforming reactions. However, as the distributed reaction regime promoted a well-mixed condition (minimizing carbon dioxide formation) and since the steam reforming reactions are considered up to three times faster [37], we conjecture that the distributed reaction regime primarily will influenced be by steam reforming reactions. Dry reforming was still thought to have occurred, but was not as active as steam reforming reactions. In conventional partial oxidation, steam reforming reactions only occurred toward the rear (tail end) of the reactor, wherein distributed reforming, steam reforming reactions were active throughout the reactor.

The experimental data conducted by this work support these observations provided here. In a low temperature reactor (700–800 °C), outlined in the works of Scenna and Gupta [1,3], the reformate consisted of low concentrations of hydrogen and significant hydrocarbon formation. From the literature, reactor temperatures lower than 800 °C limit the activity of the steam and dry reforming reactions at these temperatures. Even though the reactor temperatures limited the activity of these reactions, to result in low syngas yields, the entrainment of steam and carbon dioxide influenced the activity of the HACA soot formation mechanism. This in turn affected the hydrocarbon product distribution in the syngas. In Scenna and Gupta [1,3], it was observed that for a given O/C ratio, conditions that occurred within the flamelets in eddies regime exhibited higher concentrations of hydrogen, acetylene, and soot, which indicated activity of the HACA chemical reactions. For conditions that occurred in the distributed reaction regime, the product distribution shifted to favor higher concentrations of ethylene and methane, without observable amounts of soot formation, indicating a reduction in the HACA mechanism activity. As predicted under distributed reaction regime, hydrocarbons exhibited a higher H/C ratio as hydrogen abstraction reaction was suppressed. This can be observed from Figs. 4 and 5.

At high reactor temperatures (800–1000 °C), conditions were sufficient to activate dry and steam reforming reactions to generate higher hydrogen yields. The more distributed conditions reduced the activity in the partial oxidation reaction, which allowed for greater conversion through the steam and dry reactions. Within the high temperature work of Scenna and Gupta [2], reducing the air preheats promoted a greater distributed reactor configuration, which correlated with increased conversion and efficiency. Decrease in air preheats increased hydrogen and carbon monoxide concentrations. Under wet partial oxidation conditions, Scenna and Gupta [4] showed that addition of trace amounts of steam (S/C = 0.01) resulted in a more distributed condition and increased conversion (4.83–8.63 ± 2.1%). Steam itself reduces the chemical time and length scales, allowing a more distributed condition to emerge. However, assuming that all added steam promoted steam reforming reactions, the small amount of added steam could have only improved conversion by at most 1.53–1.59% ± 0.03%. This discrepancy in conversion was attributed to the steam reforming induced by the entrainment of hot reactive products into the fuel–air mixture. In Scenna and Gupta [55], the effects of distributed reaction regime were less discernable, as there were two competing effects when the oxygen content was varied. Increases in oxygen content resulted in a less distributed condition, but showed improved reformate quality. However, as the reactions were limited by the availability of oxygen, the addition of oxygen enhanced the extent of reforming reactions. This promoted increased conversion, which offset the negative effects caused by the reactor becoming less distributed, and any dry or steam reforming reactions that may have been promoted. In this case, the more distributed condition also promoted less steam formation, minimizing the effect of added steam reforming reactions. Figures 6 and 7 show a noticeable change in reformate quality between the high and low temperature reactors. Both reactors operated under dry partial oxidation conditions, at O/C ratios of 1.0–1.3. Higher temperature conditions improved reformate composition, which also showed increased efficiency and conversion. Higher temperature conditions were associated with greater activity in the steam and dry reforming reactions.

The distributed reaction regime and the corresponding entrainment of hot reactive products influenced the chemistry occurring within the reactor. As the current understanding of the chemistry of homogeneous reforming of larger hydrocarbons is still developing, literature was used to estimate the activity of the reforming reactions.

The entrainment of the hot reactive products (heat, steam, carbon dioxide, and other gas phase species) into the fuel–air mixture reduced the oxygen concentrations, which allowed the distributed reaction regime to emerge. These reactive species promoted the potential for steam reforming reactions and, to a lesser extent, dry reforming reactions to occur. In addition, the entrainment of steam and carbon dioxide will also suppress the formation of soot and acetylene. Steam and carbon dioxide will absorb hydrogen radicals, which prevent them from promoting hydrogen abstraction reactions that promote the formation of acetylene and soot (via HACA mechanism). The more distributed conditions allowed for greater entrainment to promote these benefits. The partial oxidation reactions produce radicals, which promote secondary dry and steam reforming reactions.

This works was funded through the United States Army's In-House Laboratory Innovative Research Program.

• U.S. Army Communications-Electronics Research, Development and Engineering Center (ILIR).

Conversion =

$molarflowrateofCO+CO2molarflowrateofcarboninfuel$

HACA =

hydrogen abstraction carbon addition

O/C =

molar oxygen to carbon ratio

S/C =

molar steam to carbon ratio