- Email: [email protected]
Effects of mixture composition, dilution level and pressure on auto-ignition delay times of propane mixtures
Chemical Engineering Journal 277 (2015) 324–333
Contents lists available at ScienceDirect
Chemical Engineering Journal journal homepage: www.elsevier.com/locate/cej
Effects of mixture composition, dilution level and pressure on auto-ignition delay times of propane mixtures P. Sabia a,⇑, M. de Joannon a, G. Sorrentino b, P. Giudicianni b, R. Ragucci a a b
Istituto di Ricerche sulla Combustione – C.N.R., Naples, Italy DICMAPI – Università Federico II, Naples, Italy
h i g h l i g h t s The ignition process of propane mixtures is investigated at intermediate temperatures. The activation energy of the ignition process changes from intermediate to high temperatures. Such behavior strongly depends on the mixture dilution levels, stoichiometry and system pressure. The kinetics of the ignition process is numerically exploited. The competition between oxidative and recombination/pyrolytic routes promotes this behavior.
a r t i c l e
i n f o
Article history: Received 16 December 2014 Received in revised form 24 April 2015 Accepted 27 April 2015 Available online 5 May 2015 Keywords: Chemical kinetics Ignition delay times Numerical simulations MILD combustion Intermediate temperatures
a b s t r a c t The auto-ignition process of propane/oxygen mixtures has been widely studied in several facilities. Literature on shock tubes and tubular flow reactors has shown a change in the activation energy of the ignition process in the transition region from intermediate to high-temperature chemistry. Although fuel ignition chemistry has been widely exploited at low and high temperatures, further studies are required at intermediate temperatures. Based on previously published experimental results, this paper aimed to investigate the kinetics responsible for such a phenomenology, through a detailed kinetic analysis of the main pathways involved in the ignition process of propane mixtures. Simulations were performed over a wide range of temperatures (from low to high), changing the dilution levels of the mixture from ‘‘air’’ conditions to 97%, the pressure from 0.1 up to 3 MPa, and the mixture compositions from lean in fuel to rich in fuel. The analysis suggests that the phenomenology strongly depends on the mixture dilution levels and pressure. In particular, the differences in the ignition chemistry between low to intermediate temperatures and between intermediate to high temperatures is more evident for systems that are highly diluted and at low pressure. Such results are supported by data from the literature. This aspect explains why such a behavior is not reported for any experimental configurations and partially justifies the differences among the data obtained in several facilities, which is commonly addressed by the non-idealities of systems. Ó 2015 Elsevier B.V. All rights reserved.
1. Introduction Understanding the characteristic auto-ignition times and their relationship with operating conditions is vital for the successful design and dimensioning of combustion systems. This is particularly true for MILD combustion processes [1], in which highly diluted mixtures with inlet temperatures higher than that of fuel auto-ignition are used, thus giving rise to a combustion mode that
⇑ Corresponding author at: P.le Tecchio 80, 80125 Naples, Italy. Tel.: +39 0817683279; fax: +39 0812391709. E-mail address: [email protected] (P. Sabia). http://dx.doi.org/10.1016/j.cej.2015.04.143 1385-8947/Ó 2015 Elsevier B.V. All rights reserved.
is drastically different from that of traditional diffusive/deflagrative flames. In these applications, large mixture dilution levels are used to maintain working temperatures below critical values for the formation of pollutants species while high inlet temperatures are used to promote and sustain the oxidation process that occurs in homogeneous and nearly isothermal conditions [1–4]. In these conditions, the combustion regime is primarily controlled by a distributed auto-ignition process, which represents a key parameter in determining combustion behaviors [3–8]. However, high initial temperatures (above the auto-ignition value) and diluted conditions have significant effects on the chemical kinetics, affecting the auto-ignition chemistry [9–13]. Fuel nature, mixture
P. Sabia et al. / Chemical Engineering Journal 277 (2015) 324–333
composition, temperature and pressure represent the main controlling parameters of the ongoing chemical reactions in a combustion process. Many detailed studies have described the complex relationships of these parameters and their respective roles in determining combustion behavior. However, reports on these parameters under MILD conditions are scarce, and many chemical kinetics effects that are specific to these conditions must be further clarified. This information is essential for developing predictive, effective and robust numerical models for practical device design. Combined experimental and modeling efforts are required to elucidate the auto-ignition dependence on the operating parameters and to develop robust and predictable chemical kinetics models of MILD combustion processes. Many experimental and model aspects must be fixed to ensure the meaningfulness of obtained results. For instance, on the experimental side, the ignition process of hydrocarbons has been extensively studied over a wide range of operating conditions (pressure, temperature and mixture composition) and in many experimental facilities. Shock tubes [14–20], flow reactors [10,21–30] and rapid compression machines (RCM) [31,32] are commonly used to evaluate the ignition process. Although each of these methodologies has its merits, their utility is restricted to well-defined ranges of pressure, temperature and ignition times. Therefore, several experimental approaches have to be pursued in parallel. By combining data from different sources, the auto-ignition process of fuel/oxygen mixtures over a wide range of temperatures and pressures can be investigated. However, comparisons among data obtained through different experimental configurations or empirical/numerical methods must be made carefully. Interesting and even unexpected behaviors can emerge when the ignition process is experimentally studied under non-conventional conditions. Sabia et al. [10] numerically investigated methane oxidation under MILD conditions. They identified a phenomenology resembling ‘‘NTC’’ behavior of methane under MILD conditions at intermediate temperatures, and they identified the kinetics responsible for the behavior [9]. This phenomenology was previously experimentally identified by Huang et al. [20] in shock tubes for lean to rich methane/air mixtures but under different reference conditions, namely at high pressures (1.6–4.0 MPa) and at intermediate temperatures. Similarly, we previously [29] studied the propane auto-ignition process in a tubular flow reactor for highly diluted mixtures that were pre-heated to a wide range of temperatures. For mixtures characterized by fuel/oxygen ratios close to the stoichiometric conditions and diluted in N2 at 90%, at intermediate temperatures auto-ignition, delay times were almost independent of the inlet temperature. For high temperatures (Tin > 1100 K), the delay times linearly changed with the inlet temperature according to the Arrhenius plot. Previous reports of ignition delays from TFRs that obtained data for stoichiometric propane/air mixtures at several pressures have indicated a linear dependency on ignition delay temperature. Other recent studies [14–19] on propane auto-ignition in shock tubes have reported a sudden change in the activation energy when temperatures shift from intermediate to high at elevated pressures. In particular, Cadman et al. [14] measured propane ignition delay times at elevated pressures (0.5–4 MPa) for temperatures in the range of 850–1100 K for fuel-lean and stoichiometric mixtures in a shock tube. They compared the acquired ignition delay times with previous high temperature data provided by Burcart et al. [15,16], identifying a decrease in the ignition process activation energy in the range of 850 K < T < 1100 K. A sensitivity analysis demonstrated the importance of hydroperoxyl, propyl and methyl radicals at these intermediates temperatures. Similar
325
considerations were made by Herzler et al. [17]. They measured the ignition delay times of lean propane–air mixtures in the temperature range of 740–1300 K at pressures of 1 and 3 MPa in a shock tube facility. The experiments confirmed that the activation energy of the ignition delay time decreases at approximately 1050 K. Zhukov et al. [18] measured the ignition delay times behind reflected shock waves for lean propane–air mixtures in the temperature range of 800–1500 K and in the pressure range of 0.2– 50 MPa. They identified two different activation energies for the auto-ignition process that could be expressed as Arrhenius-dependent, at intermediate and high temperature ranges. They also updated a detailed kinetic mechanism based on their experimental data. Penyazkov et al. [19] measured the ignition delay times for lean, stoichiometric, and rich propane–air mixtures within the temperature range of 1000–1800 K at a pressure range of 0.2–2.0 MPa. They identified two different slopes for temperatures lower and higher than 1300 K. Healy et al. [31] performed several experiments in an RCM facility of mixtures composed of different proportions of methane/propane and methane/ethane/propane and investigated the effect of pressure on the ignition delay time. They compared their results with shock tube data obtained at high temperatures, observing a marked change in ‘‘activation energy’’ as a function of temperature in the transition region from intermediate- to high-temperature chemistry. In summary, propane reactivity is sensitive to changes from intermediate to high temperatures. This property is of interest regarding the ignition chemistry of hydrocarbons because propane exhibits thermo-chemical and combustion properties of larger hydrocarbon fuels more closely than either methane or ethane. Furthermore, propane oxidation is a sub-scheme of larger hydrocarbons, and thus understanding the ignition chemistry of such a reference fuel can provide useful information for a larger class of fuels. According to the previous considerations, the present work aimed to better define the oxidation chemistry of propane auto-ignition delay times by utilizing an extensive simulation of the chemical kinetics and comparing the numerical results with experimental data reported in the literature for several facilities and over a large range of operating conditions. The study focuses on the effects of the fuel/oxygen ratio, mixture dilution level and pressure at low, intermediate and high (750–1400 K) inlet temperature ranges. The objective is to identify possible chemical pathways responsible for the different auto-ignition trends observed across the various temperature ranges.
2. Critical analyses of test facilities Prior to analyzing the effects of mixture composition, dilution and pressure on the ignition data, a critical review on the reliability of ignition data is mandatory because non-idealities and perturbations may alter the data quality obtained from the reference facilities (shock tubes, tubular flow reactors and RCMs). A shock tube uses the compressive heating of a shock wave to rapidly increase the temperature and pressure of a premixed combustible mixture. Shock tubes are used to study the ignition process within short periods of times (on the order of tens of microseconds). Problems related to shock tubes have been widely discussed in the literature [32–35]. Recently, Davidson and Hanson [34] provided a clear and exhaustive analysis of shock tube performances. They identified the formation of a viscous boundary layer behind incident and reflected shock waves as a perturbing effect. This non-ideality causes a temperature and pressure increase in the core of the flow and attenuation of an incident shock wave,
326
P. Sabia et al. / Chemical Engineering Journal 277 (2015) 324–333
which significantly affects the apparent activation energy of the ignition process. Experiments are performed in argon, rather than nitrogen, to eliminate the possible influence of vibrational relaxation in the carrier gas and to eliminate or minimize shock wave bifurcation near the wall in the reflected shock front. Reflected shock temperatures are higher than those for nitrogen, influencing the ignition process timing history. Therefore, differences between combustion data in air and argon should be recognized and quantified. Furthermore the constant volume modeling approach is not valid because the energy released during the ignition process causes a rarefaction of the local pressure [35]. Such a condition is more stringent for highly concentrated fuel/oxygen mixtures for which the heat release is high. RCMs operate by compressing a homogeneous fuel/oxidizer mixture to moderate temperatures (up to Tmax 1200 K) and high pressures (Pmax 7 MPa) in a cylinder via the motion of a piston. The temperature and pressure can be sustained for longer than 10 ms. RCMs reproduce the processes that take place in devices such as spark engines and homogeneous charge compression ignition (HCCI) engines. Nevertheless, the time required to compress the test gas mixtures limits the minimum characteristic time of the investigation to 1 ms. One difficulty related to RCM operation is producing spatially homogenous conditions. Boundary layers, roll-up vortexes and heat loss to the walls during mixture compression can potentially change the temperature field in the cylinder and affect the measured ignition delay time [36–41]. Some strategies have been adopted [31,41] to avoid the formation of roll-up vortexes, but various non-idealities still remain. In a flow reactor (TFR), a pre-heated carrier gas flow rapidly mixes with a fuel flow, and the premixed charge enters a duct. The mixture chemically reacts as it flows downstream through the duct. Typical ignition times are on the order of milliseconds. The pre-heating temperatures are limited to electric heaters, and the pressures are moderate. To obtain useful kinetics data from flow reactor experiments, many of the secondary effects, such heat loss and species dispersion/diffusion, must be rendered negligible [22,23,27,28,42]. Furthermore, the mixing times must be shorter than the chemical kinetic times to insure true premixed conditions [42,43] at the entrance of the system. System non-idealities cannot be easily overcome, but the knowledge of their influence on data quality must be carefully addressed to interpret ignition data. The reference facilities may function properly under different operating conditions regarding pressure, inlet temperature and the range of ignition delay times. By combining data from these several facilities, it is possible to investigate the auto-ignition process of fuel/oxygen mixtures over a wide range of conditions.
3. Propane auto-ignition delay data Fig. 1 shows an Arrhenius plot of propane auto-ignition times (tign.) measured in several facilities that depicts stoichiometric conditions versus reactant inlet temperature (Tin). Although the figure does not show all available auto-ignition delay data on propane mixtures, it highlights the temperature dependence of various auto-ignition behaviors in several experimental configurations under different operating conditions. They are summarized in the upper part of Fig. 1. The ignition delay times are relative to a propane stoichiometric mixture in air conditions (approximately 76% in N2), except the data from Sabia et al. [29] where the mixture dilution level is 90% in N2, resembling MILD [1] conditions. Auto-ignition delay data in a RCM facility [31] show the typical NTC behavior of propane/air mixtures at low temperatures and high pressures. This behavior is not extensively discussed in the
Fig. 1. Literature auto-ignition delay data from several facilities at different operating conditions.
paper, but it was reported for completeness of the ignition phenomenologies related to propane mixtures. At higher inlet temperatures, data are generally obtained in TFR reactors. Fig. 1 shows data for a propane/oxygen stoichiometric mixture at several pressures, namely 0.1 MPa [22,26,27,29], 0.2 MPa [21], 0.9 MPa [28] and 1 MPa [25]. Under such operating conditions data show a linear trend with Tin, except the data by Holton et al. [27] and by Sabia et al. [29]. In such cases, the auto-ignition delay time curve shows two different slopes. In particular the data from [27] showed a linear trend of tign. with the inlet temperature for 1000/Tin > 1 in the Arrhenius plot, while for 1000/Tin < 1 the slope of the curve slightly diminishes. Sabia et al. [29] showed that for inlet temperatures lower than approximately 1100 K (1000/Tin = 0.9) tign. is almost independent on Tin, whereas for 1000/Tin < 0.9 it diminishes linearly with temperature. A similar behavior was reported in the literature for data obtained using shock tube facilities. In particular, Cadman et al. [14] identified a decrease in the ignition process activation energy in the range 850 < Tin < 1100 K compared to the high temperature auto-ignition values provided by Burcart et al. [15] at 0.5 MPa. Penyazkov et al. [19] also experimentally observed a change in the slope of the tign. vs Tin profiles at approximately 1300 K in the pressure range 1.2–2.0 MPa. These reports, obtained for stoichiometric propane mixtures in air at several pressures, suggest that the change in the ignition chemistry temperature also depends on pressure. Furthermore, results by Sabia et al. [29] indicated that highly diluted mixtures are more prone to show a change in slope between low and higher temperatures than undiluted (‘‘air’’) conditions at low pressures. These results, confirmed by those collected for fuel-lean mixtures in shock tube facilities [14,15,17] supports the idea that mixture dilution level also may play a role in the onset of different ignition regimes passing from low to intermediate and high temperatures. Sabia et al. [29] found experimentally that the change in the activation energy of propane ignition at intermediate temperatures depends also on the stoichiometry of the mixtures. In particular,
P. Sabia et al. / Chemical Engineering Journal 277 (2015) 324–333
they identified such a behavior from fuel-lean to rich mixtures in the range of 0.15 < C/O < 0.35 and mixture inlet temperatures from 975 to 1100 K. The variation of the activation energy of the autoignition process passing from intermediate to high temperatures for propane mixtures was mainly evidenced in studies realized in shock tubes under low-pressure conditions. In these operating conditions, measurements are most likely affected by the insurgence of non-ideal effects. Therefore, the auto-ignition dependence on the temperature was quite often attributed to these non-idealities [30,32,33,44]. On the other hand, the experiments by Holton et al. [27] and Sabia et al. [29] realized in tubular flow reactors have confirmed that propane mixtures exhibit a change of the activation energy of the auto-ignition process at intermediate temperatures where the non-linear effects can be neglected. A plausible alternative explanation of the observed behavior could be a change in the chemistry of propane mixtures leading to the ignition process. While the ‘‘NTC’’ behavior at low temperature has been widely investigated and the kinetics responsible of such a behavior completely identified, in literature there is not a thorough analysis of the phenomenologies observed at intermediate temperatures. The data from literature suggest that the change in the functional dependence of the auto-ignition on inlet temperatures from intermediate to high temperatures is complexly dependent on pressure (with mutual interactions), dilution level and mixture composition. The analysis of such a dependence thus requires a particular care and a thorough exploration of the variation of these parameters taking in consideration their possible interactions. 4. Numerical tools Simulations of the propane auto-ignition process in a plug-flow reactor were carried out using the PLUG application of ChemKin 3.7 [45] under adiabatic conditions. The absence of heat loss mechanisms from the reactor to the surroundings allows for the evaluation of only the kinetic aspects of the problem without considering complex interactions between heat exchange mechanisms and chemical reactions [10,11]. In agreement with previous studies [10,29], the mechanism by CRECK Modeling Group at Politecnico di Milano [46] was used to perform the simulations. It was validated in a wide range of operating conditions, from low temperatures, through the NTC behavior, up to high temperatures and pressures [47–49]. Before performing simulations, the kinetic mechanism was tested over a wide range of temperatures and pressures considering data from RCM, TFR and shock tube facilities. Fig. 2 shows a comparison among simulations and experimental data from the literature. Different criteria were applied to value the trustability of the kinetic mechanism on the basis of experimental
327
facility features and analytical procedure to value the ignition times. The RCM and shock tube auto-ignition delay times were simulated using the AURORA application of ChemKin 3.7 [45] and were considered as closed constant volume systems under adiabatic conditions. For such systems auto-ignition delay times were estimated on the basis of the temperature inflection point of the temporal temperature profile. The data from the tubular flow reactors were simulated using the PLUG application of ChemKin 3.7, considering no heat loss from the reactor to the surroundings, while the ignition criterion was based on a specified temperature increase beyond the initial temperature. In the case of TFR systems, the autoignition delay time was defined on the basis of the temperature profiles, considering as reference time the one corresponding to the abrupt temperature increase in air condition, while, in the case of diluted systems [29] a temperature increase beyond the inlet one of 10 K, according to previous studies [9,10,29]. In MILD combustion processes a criterion ad hoc has to be chosen [50]. Namely, under such operating conditions, the temperature increase during the oxidation process condition is relatively low because of high-diluted levels, and the temperature increase is not abrupt as in traditional flame conditions. Especially under ‘‘low reactive conditions’’, the temperature increase is very modest and the assumption of high temperature increase to define the ignition time would cause a net loss of information and the failure in identifying peculiar oxidation regimes. Considering the TFR system and the oxidation process features, we adopted a gradient of 10 K for the estimation of the auto-ignition delay time. The data reported by Beerer et al. [28] are relative to fuel-lean mixtures at 0.9 MPa, whereas the data reported by Sabia et al. [29] consider stoichiometric propane/oxygen mixtures diluted at 90% in N2 at environmental pressure. The ignition delay times by Penyazkov et al. [19] were obtained in a shock tube at 1.2– 2.0 MPa, whereas the data obtained by Cadman et al. [14] are relative to a lean propane/oxygen mixture (phi = 0.5) diluted in Ar at 0.5 MPa. The last data are different with respect to the one reported in Fig. 1 to value also the robustness of the kinetic scheme varying the mixture carbon/feed ratio. For low temperatures, the simulated data are relative to the RCM by Gallagher et al. [32] at 3 MPa. In such cases, the detailed mechanism provides a good approximation of the experimental data and is able to reproduce the NTC behavior. In such a region, the differences between numerical and experimental data are likely due to heat exchange effects that were not considered in the simulations. For TFR and shock tube data, the mechanism is able to reasonably reproduce both the auto-ignition delay time data and the slopes of curves. Some considerations apply for the data from shock tube facilities for both stoichiometric and fuel lean mixtures. In this context, it is worth noting that Sabia et al. [10,29] showed some discrepancies between their numerical and experimental data for fuel rich mixtures but the kinetic schemes well reproduced the dependence of the ignition delay time on the variation of this parameter. In general, under any condition, there is a reasonable agreement between simulations and experimental ignition delay times in the literature.
5. Numerical results
Fig. 2. Comparison among auto-ignition delay and simulation data reported in the literature.
The simulations were realized using the PLUG application of the ChemKin 3.7 software. In the next paragraphs, the auto-ignition delay times, the calculated for mixtures with different fuel/oxygen ratios, the mixture dilution levels and the pressures are reported.
328
P. Sabia et al. / Chemical Engineering Journal 277 (2015) 324–333
The analyses were performed primarily considering the operating conditions reported in Sabia et al. [29], namely the diluted conditions. The analyses are parametrically extended to ‘‘air’’ condition or to high pressures. This approach was chosen because under diluted conditions the kinetics is relatively slower with respect to ‘‘air’’ conditions and the different pathways involved in the auto-ignition process are easier discernable. This allows understanding the effects of the investigated parameters on the auto-ignition chemistry in a clearer way. 5.1. Effect of fuel/oxygen ratio Fig. 3 shows the auto-ignition delay times in the Arrhenius plot computed for fuel-rich, stoichiometric and fuel-lean conditions identified using C/O = 0.6, 0.3 and 0.05, respectively, at inlet temperatures from 800 to 1400 K at atmospheric pressure. The mixtures are diluted at 90% in N2. The figure shows that for low inlet temperatures, the auto-ignition delay times decrease from fuel-lean to fuel-rich mixtures. At temperatures below 1125 K (1000/Tin = 0.88), for the fuel-lean mixture, the auto-ignition delay values exhibit a linear dependence on temperature in the Arrhenius diagram plot. For the fuel stoichiometric and fuel-rich mixtures, there is a change in the slope of the auto-ignition times curves at approximately Tin = 950 K. Then, as for 1000/Tin = 0.8, the slopes of the curves change again, and for high inlet temperatures, fuel-lean mixtures correspond to lower ignition times. For high temperatures, the fuel-lean mixture exhibits a second change in slope, and thus the auto-ignition data shifts toward the stoichiometric curves. This behavior was experimentally determined by Qin et al. [51].
5.2. Effect of dilution levels Fig. 4 reports the auto-ignition time profiles in the Arrhenius plane for increasing dilutions of the stoichiometric mixtures with nitrogen, corresponding to standard ‘‘air’’ conditions (approximately 76%) to 97% nitrogen. The pressure is atmospheric. The auto-ignition delay times increase from a dilution level of 76% (air) to 97%. A change in slope is observed for all dilution values. It is worth noting that it becomes more evident as the nitrogen concentration increases. For instance, under ‘‘air’’ conditions, the change in the activation energy of the ignition process is almost unperceivable. Furthermore, the ignition profile inflection point shifts towards lower inlet temperatures with increasing mixture dilutions. In ‘‘air’’ conditions, the ignition process is sustained by high temperature gradients, and a fast chemistry dominates the process evolution. In contrast, for diluted conditions, the reactivity of the system slows because lower adiabatic flame temperatures make the system more sensitive to the operating conditions and enhances the competition among several pathways [29], thus permitting the onset of phenomenologies that are generally hidden during conventional combustion processes. 5.3. Effect of pressure The same simulations were performed for various pressures. Fig. 5 shows the numerical results for a stoichiometric mixture diluted in N2 at 90% at pressures ranging from 0.1 MPa to 3.0 MPa. The temperature range was extended to include low temperatures under the RCM conditions. In agreement with data from the literature, auto-ignition delay times decrease with increasing pressure. The slope variation between intermediate to high temperatures is less evident as the system pressure is increased and is shifted to higher inlet temperatures. In addition, the typical low temperature NTC behavior is observed at 2.0 MPa and becomes evident at 3.0 MPa over the same temperature range identified by Gallagher et al. [32]. 6. Discussion
Fig. 3. Auto-ignition delay data for fuel-lean, stoichiometric and fuel-rich conditions.
Fig. 4. Auto-ignition delay data for stoichiometric mixtures at 0.1 MPa with different dilution levels.
The numerical results reported in the previous section highlight unique behaviors related to the auto-ignition delay times of propane mixtures. In particular, the Arrhenius diagram plot of the auto-ignition times for intermediate temperatures exhibited a change in slope, which was dependent on the fuel/oxygen feed ratio, the mixture dilution level and the pressure. These results are supported by literature experimental data obtained from several facilities and operating conditions (Fig. 1). It is of uttermost interest to identify in the critical chemical kinetics pathways that
Fig. 5. Auto-ignition delay data for a stoichiometric mixture at several pressures.
329
P. Sabia et al. / Chemical Engineering Journal 277 (2015) 324–333
lead to the observed phenomenology. To this aim, standard chemical kinetics numerical tools, such as flux diagrams, and sensitivity and rate of production analyses of main species, are commonly used. In the following subsections, these methods will be presented, referencing the cases discussed in the previous section. 6.1. Main reactions involved in propane auto-ignition In Sabia et al. [29], the results relative to a stoichiometric mixture diluted at 90% at three different inlet temperatures, namely low, intermediate and high temperatures, at atmospheric pressure were reported. A comprehensive flux diagram is reported in Fig. 6 in order to better visualize and follow the different kinetic pathways. The flux diagram is relative to the auto-ignition time, identified on the basis of criterion earlier reported. Tree different colors identify the kinetic pathways at low (light gray), intermediate (dark gray) and high (black) temperature. Continuous lines refer to the carbon species pathways, while the dashed ones to radical production. They feed the branching mechanisms identified in the dashed-line rectangular insets. Reaction by reaction, the position of species reported aside any continuous line identifies the relative importance of reactants involved in the consumption of the carbon species reported in round frames.
At low temperatures, propane is dehydrogenated to isopropyl and normal-propyl by OH and HO2 radicals. Such C3 radicals mainly undergo dehydrogenation reactions, reacting with molecular oxygen (reactions 1 and 2 in Table 1). Propyl radicals also decompose to CH3 and C2H4 via reaction 3. Methyl radicals are oxidized to CH3O and CH2O via reactions 4–6, leading to the production of OH and H radicals. The HO2 radicals produced by reaction 1 and 2 primarily react through reactions 5 and 7. Finally, reaction 8
Table 1 Primary reactions involved in the auto-ignition process of propane mixtures. #
Reaction
1 2 3 4 5 6 7 8 9 10 11 12
O2 þ n-C3 H7 ) HO2 þ C3 H6 O2 þ i-C3 H7 ) HO2 þ C3 H6 n-C3 H7 ¼ CH3 þ C2 H4 CH3 þ CH3 OO ¼ CH3 O þ CH3 O CH3 þ HO2 ¼ CH3 O þ OH CH3 O þ M ¼ CH2 O þ H HO2 þ HO2 ¼ H2 O2 þ O2 H2 O2 ðþMÞ ¼ OH þ OHðþMÞ i-C3 H7 ¼ C3 H6 þ H CH3 þ CH3 þ M ¼ C2 H6 þ M H þ O2 ¼ OH þ O H þ O2 þ M ¼ HO2 þ M
Fig. 6. Comprehensive flux diagram at environmental pressure.
330
P. Sabia et al. / Chemical Engineering Journal 277 (2015) 324–333
represents the branching reaction at low temperature promoting the formation of OH radicals. At intermediate temperatures, while isopropyl radicals continue to react with oxygen to form HO2 radicals (reaction 1), propyl radicals, in contrast with the low temperature kinetics, primarily decompose to methyl radicals and ethylene (reaction 3). i-C3H7 also decomposes via an equilibrium reaction (reaction 9 in Table 1) that reconverts propene to i-C3H7 consuming H radicals. The branching reactions are related to H2O2 formation and decomposition, but the formation of HO2 radicals is relatively limited with respect to the previous flux diagram because n-C3H7 reacts via reaction 3 while i-C3H7 is only partially dehydrogenated by O2. The production of radical species is sustained by the pathway:
The typical high temperature branching reaction (reaction 11) becomes the branching mechanism in this temperature range and significantly increases the reactivity of the system, leading to a higher slope of auto-ignition delay time curve slope in the Arrhenius plot diagram with respect to intermediate temperatures. 6.2. Reaction rate analyses
Reaction 10 competes with the oxidation pathways, partially slowed by the relative depletion of HO2 radicals, lowering the production of OH and H radicals. This effect, combined with the less pronounced production of HO2, leads to lower reactivity of the system and to a smaller change in the slope of the auto-ignition time in the Arrhenius plot diagram. For high inlet temperatures, numerical analyses suggest that propane is dehydrogenated by OH and H radicals to normal and isopropyl species, decomposing via reactions 3 and 9, respectively. The last reaction boosts the production of H radicals. Propane also thermally decomposes to CH3 and C2H5. Methyl radicals mainly recombine to ethane feeding the pathways C2 H6 ) C2 H5 ) C2 H4 ) C2 H3 . The methyl oxidation routes are relatively less intense due to recombination reaction strong activation and to the relatively depletion of HO2 radicals.
The description of the kinetics involved in propane auto-ignition as a function of the inlet temperature is illustrated in Fig. 6, where the rate of key reactions, identified using sensitivity analyses (provided in Supplementary material) and flux diagrams, is reported as a function of the inlet temperature. In particular, Fig. 6a refers to the branching reactions, whereas Fig. 6b to C1 species. Fig. 7a describes the trend of reactions 7, 8 and 11. Furthermore, reaction 12 is reported in Table 1 because, under several operating conditions (e.g., high pressure), it competes with reaction 11. Notably, reactions 7 and 8 control the branching mechanism up to 1200 K, whereas reaction 11 is the most important reaction for higher inlet temperatures. Reaction 12 is negligible for any considered temperature at atmospheric pressure. Fig. 7b shows the relative weight of C1 oxidation and the recombination reactions. For Tin < 900 K, the C1 oxidation reactions, represented by reactions 4 and 5, are dominants. At higher temperatures, the recombination pathway (reaction 10) becomes progressively more relevant, and for Tin > 1100 K, this pathway is the dominant one. The onset of the recombination pathway slows the mixture reactivity because it favors the storage of C and H radicals in the form of C2 species. This inhibits the oxidation channel from producing radical species in a temperature range where the main branching mechanism (reaction 7 + 9) is relatively slow. This mechanism delays the auto-ignition process. At higher temperatures, reaction 11 dominates the branching mechanism, providing many radicals, accelerating the system reactivity, and inducing a change in the slope of the auto-ignition delay time profile. Similar analyses have been realized for fuel-rich and fuel-lean conditions.
Fig. 7. Reaction rate analysis for a stoichiometric C3H8/O2 mixture diluted in 90% N2 at atmospheric pressure.
Fig. 8. Reaction rate analysis for a fuel-rich (C/O = 0.6) C3H8/O2 mixture diluted in N2 at 90% at atmospheric pressure.
CH3 ) CH3 O ) CH2 O ) HCO ) CO In particular, methyl radicals are oxidized to CH3O by HO2 radicals. For Tin > 1000 K, reaction 10 begins to play an important role and promotes recombination and pyrolytic reactions, described by this sequence:
C2 H6 ) C2 H5 ) C2 H4 ) C2 H3
P. Sabia et al. / Chemical Engineering Journal 277 (2015) 324–333
Fig. 8 shows the rate of production of key reactions versus the inlet temperature relative to a rich mixture with a fuel/oxygen ratio equal to 0.6. Fig. 8a also shows the analysis relative to the rate of the key branching reactions, and Fig. 8b shows the reaction relative to C1 species. In both cases, the auto-ignition delay times versus inlet temperature are reported. As shown in Fig. 8b, for low temperatures, reaction 5 converts methyl radicals to CH3O and OH, which feed the oxidation channel:
CH3 ) CH3 O ) CH2 O ) HCO ) CO At 1000 K, the rate of reaction 5 equals that of reaction 10, thus delaying auto-ignition, as shown in the auto-ignition curve. At even higher temperatures, the recombination channel represents the main pathways of C1 species. Fig. 8a shows that the main branching mechanism up to 1200 K corresponds to reaction 8 (H2O2 (+M) = OH + OH (+M)), whereas for Tin > 1200 K, reaction 11 (H + O2 = OH + O) enhances the system reactivity and changes the slope of the auto-ignition delay time curve. For Tin > 1000 K, the rate of reaction 8 slowly increases, and for Tin > 1200 K, the reaction rate decreases due to the lack of HO2 radicals as C3H7 radicals are mainly decomposed and they do not react with O2, which leads to the production of HO2 radicals. In addition, CH3 radicals recombine to C2H6, which diminishes the production of radicals from the oxidation pathways. Fig. 9a and b show the rates of key reactions for fuel-lean mixtures. In reactions of fuel-lean mixtures, as also described in Fig. 3, the auto-ignition delay data exhibits a slight change in slope from intermediate to high temperatures. At 1100 K, the rates of reactions 5 and 10 are equal (Fig. 9a) while reaction 11 proceeds faster than reaction 8 (Fig. 9b). The delaying effect induced by reaction 10 is damped by the promotion of reaction 11, resulting in a slight change in the auto-ignition delay time curve. Furthermore, the flux diagrams and sensitivity analyses for fuel-lean mixtures (C/O = 0.05) (not reported here) have revealed that the competition between oxidation and recombination channels is altered by the promotion of other oxidative routes represented by the following reactions:
Fig. 9. Reaction rate analysis for a fuel-lean (C/O = 0.05) C3H8/O2 mixture diluted in N2 at 90% at atmospheric pressure.
331
1. CH3 + OH = CH3OH 2. CH3 + OH = CH2(S) + OH. The onset of such reactions, promoted by the high amount of OH radicals boosted by reaction 11, makes the recombination channel less active, thus lowering its effect on the ignition delay times. Furthermore, for temperatures higher than 1300 K, the considered reaction rates diminish. Such an effect is also observed in the trend of the auto-ignition delay times. The fast conversion of H to OH radicals by reaction 11 diminishes the concentration of H radicals that promptly dehydrogenate propane to its radicals at high temperatures. The rate of production analysis has been extended to stoichiometric mixtures diluted at 95% in N2 at atmospheric pressure to assess the influence of such a parameter on the kinetics of controlling the auto-ignition times. Fig. 10a and b are relative to the branching reactions and the C1 species. The reaction rate of the recombination reaction becomes higher than the oxidation one (represented by reaction 5) for Tin = 950 K. Therefore, because dilution promotes the recombination channel with respect to the oxidation routes for lower temperatures, the change in slope of the curve of the auto-ignition delay time occurs at lower temperatures with increasing dilution levels. Fig. 10a shows that the high temperature branching of reaction 11 becomes faster than reaction 8 for Tin = 1150 K, widening the range of temperatures for which reaction 8 is the primary branching reaction. Similar analyses were performed for stoichiometric mixtures diluted at 90% in N2 at 3.0 MPa. Fig. 11 shows the main kinetic reactions involved at the auto-ignition time as a function of inlet temperature for a stoichiometric mixture diluted in nitrogen at 90% at 3.0 MPa. Fig. 11b shows that the reaction CH3 + CH3OO = CH3O + CH3O mainly consumes CH3 radicals while reaction 11 becomes faster than reaction 5 at Tin = 1200 K. Therefore, the role of the recombination channel is primarily suppressed by reaction 4. Flux diagrams and sensitivity analyses showed that high pressures alter
Fig. 10. Reaction rate analysis for a stoichiometric C3H8/O2 mixture diluted in N2 at 95% at atmospheric pressure.
332
P. Sabia et al. / Chemical Engineering Journal 277 (2015) 324–333
Fig. 11. Reaction rate analysis for a stoichiometric C3H8/O2 mixture diluted in N2 at 90% at 3.0 MPa.
the equilibrium of reaction CH3 + O2 = CH3OO, producing more methyl peroxide radicals that in turn, participate in reaction 4 and boost the oxidation channel route. At high pressures, the fall-off reaction 10 (CH3 + CH3 + M = C2H6 + M) becomes bimolecular such that the dependence on the mixture dilution level becomes negligible, in contrast to diluted and low pressure systems. Fig. 11a shows that only at Tin = 1400 K, the rate of production of reaction 11 becomes higher than that of reaction 8. In addition, pressure promotes the competing reaction 12 that feeds reactions 7 and 8. The promotion of such kinetic pathways is more evident at higher pressures. Such consideration implies that the change in slope of the auto-ignition delay time curve is less evident and shifts to higher inlet temperatures when the pressure is increased. Previous reports have discussed the role of C1 recombination and oxidation routes and the competition among branching mechanisms. In addition, the prevailing kinetic route is strongly dependent on the C/O feed ratio, pressure and dilution levels. These parameters influence the relative production and distribution of radical species and the C1 oxidation/recombination competition mechanisms previously described.
7. Conclusions The auto-ignition process of propane/oxygen mixtures was numerically investigated with respect to mixture composition (carbon/oxygen ratio), dilution levels and pressure. Available literature data acquired using TFRs, shock tubes and RCMs have revealed a marked change in the ‘‘activation energy’’ of the auto-ignition process as a function of temperature in the region of transition from intermediate- to high-temperature chemistry. Such a behavior was reported in shock tubes for propane mixtures under low pressures and from intermediate to high temperatures. These operating conditions are not ideal for shock tube facilities, thus this behavior was mainly justified on the basis of non-idealities occurring in shock tube facilities exercised under these operating conditions.
Sabia et al. [29] reported the same behavior for diluted propane mixtures in a tubular flow reactor. Holton et al. [27] showed some other evidences of the insurgence of this behavior in a tubular flow reactor. An extensive kinetic analysis of this behavior has not been reported in literature. Thus the paper was devoted to the identification of the chemical pathways responsible of the onset of this phenomenology. Numerical results showed that this behavior is dependent on the fuel/oxygen feed ratio, mixture dilution levels and pressure confirming results from literature experimental data acquired from several facilities and under many operating conditions. The changes in auto-ignition times due to inlet temperature from intermediate to high temperatures depend on the operative parameters in addition to mutual, complex interactions. Therefore, propane reactivity is altered corresponding to intermediate versus high temperature regions. To identify the possible kinetic effects responsible for this phenomenology, a thorough numerical simulation analysis was undertaken here. The results of this analysis, as reported in the present study, show that the ignition chemistry is very sensitive to the competition between C1 oxidation and recombination/pyrolytic routes. Pressure, mixture dilution levels and composition alter the competition and thus promote different pathways. These parameters affect the relative production and distribution of radical species and the competition between C1 oxidation/recombination mechanisms.
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.cej.2015.04.143.
References [1] A. Cavaliere, M. de Joannon, Prog. Energy Combust. Sci. 30 (4) (2004) 329–366. [2] M. de Joannon, A. Saponaro, A. Cavaliere, M. de Joannon, Proc. Combust. Inst. 28 (2) (2000) 1639–1646. [3] I.B. Özdemir, N. Peters, Exp. Fluids 30 (2001) 683–695. [4] G.G. Szegö, B.B. Dally, G.J. Nathan, Combust. Flame 156 (2009) 429–438. [5] E. Oldenhof, M.J. Tummers, E.H. van Veen, D.J.E.M. Roekaerts, Combust. Flame 158 (8) (2011) 1553–1563. [6] M. de Joannon, G. Sorrentino, A. Cavaliere, Combust. Flame 159 (5) (2012) 1832–1839. [7] T. Plessing, N. Peters, J.G. Wünning, Proc. Combust. Inst. 27 (2) (1998) 3197– 3204. [8] C. Duwig, B. Li, Z.S. Li, M. Aldén, Combust. Flame 159 (1) (2012) 306–316. [9] P. Sabia, M. de Joannon, A. Picarelli, A. Chinnici, R. Ragucci, Fuel 91 (1) (2012) 238–245. [10] P. Sabia, M. de Joannon, A. Picarelli, A. Chinnici, R. Ragucci, Combust. Flame 160 (1) (2013) 47–55. [11] M. de Joannon, A. Cavaliere, T. Faravelli, E. Ranzi, P. Sabia, A. Tregrossi, Proc. Combust. Inst. 30 (2) (2005) 2605–2612. [12] A. Parente, C. Galletti, L. Tognotti, Int. J. Hydrogen Energy 33 (2008) 7553– 7564. [13] J. Aminian, C. Galletti, S. Shahhosseini, L. Tognotti, Flow Turbul. Combust. 88 (2012) 597–623. [14] P. Cadman, O.T. Geraint, P. Bulter, Phys. Chem. Chem. Phys. 2 (2000) 5411– 5419. [15] A. Burcat, K. Scheller, A. Lifshitz, Combust. Flame 16 (1) (1971) 29–33. [16] A. Burcat, K. Lifshitz, K. Scheller, G.B. Skinner, Proc. Combust. Inst. (1971) 745. [17] J. Herzler, L. Jerig, P. Roth, Combust. Sci. Technol. 176 (10) (2004) 1627–1637. [18] V.P. Zhukov, V.A. Sechenov, A.Yu. Starikovskii, Kinet. Catal. 46 (3) (2005) 319– 327. [19] O.G. Penyazkov, K.A. Ragotner, A.J. Dean, B. Varatharajan, Proc. Combust. Inst. 30 (2) (2005) 1941–1947. [20] J. Huang, W. Bushe, Combust. Flame 144 (1–2) (2006) 74–88. [21] V.Ya. Basevich, A.A. Borisov, S.M. Frolov, K.Ya. Troshin, G.I. Skachkov, in: Proceedings of the Fifth International Colloquium on Pulsed and Continuous Detonation, Moscow, Russia, 2006. [22] W. Freeman, A. Lefebvre, Combust. Flame 58 (1984) 153–164. 25. [23] A. Lefebvre, W. Freeman, L. Cowell, 1986, NASA CR-1750. [24] E. Lezberg, NACA TN-4028. 27, 1957. [25] L. Cowell, A. Lefebvre, SAE Technical Paper 860068, 1986.
P. Sabia et al. / Chemical Engineering Journal 277 (2015) 324–333 [26] C.J. Chang, A. Thompson, R. Winship, Ignition delay of propane in air between 725–800 °C under isothermal conditions, in: 7th Symposium (International) on Combustion, Academic Press, New York, 29 (1959) 431–435. [27] M.M. Holton, G.S. Jackson, P. Gokulakrihnan, M.S. Klassen, R.J. Roby, J. Eng. Gas Turbines Power 132 (9) (2010). [28] J. Beerer, V.G. McDonell, Proc. Combust. Inst. 33 (2011) 301–307. [29] P. Sabia, M. de Joannon, M. Lubrano Lavadera, P. Giudicianni, R. Ragucci, Combust. Flame 161 (12) (2014) 3022–3030. [30] A. Schönborn, P. Sayad, A.A. Konnov, J. Klingmann, Combust. Flame 160 (6) (2013) 1033–1043. [31] D. Healy, H.J. Curran, S. Dooley, J.M. Simmie, D.M. Kalitan, E.L. Petersen, G. Bourque, Combust. Flame 155 (3) (2008) 451–461. [32] S.M. Gallagher, H.J. Curran, W.K. Metcalfe, D. Healy, J.M. Simmie, G. Bourque, Combust. Flame 153 (1) (2008) 316–333. [33] L. Petersen, R.K. Hanson, Shock Waves 10 (2001) 405–420. [34] D.F. Davidson, R.K. Hanson, Int. J. Chem. Kinet. 36 (9) (2004) 510–523. [35] M. Chaos, F.L. Dryer, Int. J. Chem. Kinet. 42 (3) (2010) 143–150. [36] R. Minetti, M. Carlier, M. Ribaucour, E. Therssen, L.R. Sochet, Combust. Flame 102 (3) (1995) 298–309. [37] G. Mittal, M. Chomier, Combust. Flame 161 (2014) 75–83. [38] J. Clarkson, J.F. Griffiths, J.P. Macnamara, B.J. Whiteaker, Combust. Flame 125 (3) (2001) 1162–1175. [39] C.J. Sung, H.J. Curran, Prog. Energy Combust. Sci. 44 (2014) 1–18.
[40] [41] [42] [43] [44] [45]
[46] [47] [48]
[49] [50] [51]
333
G. Mittal, C.J. Sung, Combust. Flame 145 (2006) 160–180. G. Mittal, A. Bhari, Combust. Flame 160 (2013) 2975–2981. H. Mohamed, B. Hmaeid, Int. J. Energy Environ. Eng. 3 (22) (2012). H. Wu, M. Ihme, Combust. Flame 161 (2014) 2317–2326. K.-Y. Lam, Z. Hong, D.F. Davidson, R.K. Hanson, Proc. Combust. Inst. 33 (2011) 251–258. R.J. Kee, J.A. Miller, M.E. Coltrin, J.F. Grcar, E. Meeks, H.K. Moffat, A.E. Lutz, G. Dixon-Lewis, M.D. Smooke, J. Warnatz, G.H. Evans, R.S. Larson, R.E. Mitchell, L.R. Petzold, W.C. Reynolds, M. Caracotsios, W.E. Stewart, P. Glarborg, C. Wang, O. Adigun, W.G. Houf, C.P. Chou, S.F. Miller, Chemkin Collection, Release 3.7.1, Reaction Design, San Diego, CA, 2003. E. Ranzi, A. Frassoldati, R. Grana, A. Cuoci, T. Faravelli, A.P. Kelley, C.K. Law, Prog. Energy Combust. Sci. 38 (4) (2012) 468–501. E. Ranzi, T. Faravelli, P. Gaffuri, G.C. Pennati, A. Sogaro, Combust. Sci. Technol. 100 (1–6) (1994) 299–330. E. Ranzi, C. Cavallotti, A. Cuoci, A. Frassoldati, M. Pelucchi, T. Faravelli, Combust. Flame (2015), http://dx.doi.org/10.1016/j.combustflame.2014.11.030 (article in press). E. Ranzi, A. Frassoldati, A. Stagni, M. Pelucchi, A. Cuoci, T. Faravelli, Int. J. Chem. Kinet. 46 (9) (2014) 512–542. M. de Joannon, A. Cavaliere, R. Donnarumma, R. Ragucci, Proc. Combust. Inst. 29 (1) (2002) 1139–1146. Z. Qin, H. Yang, W.C. Gardiner, Combust. Flame 124 (2001) 246–254.
Contents lists available at ScienceDirect
Chemical Engineering Journal journal homepage: www.elsevier.com/locate/cej
Effects of mixture composition, dilution level and pressure on auto-ignition delay times of propane mixtures P. Sabia a,⇑, M. de Joannon a, G. Sorrentino b, P. Giudicianni b, R. Ragucci a a b
Istituto di Ricerche sulla Combustione – C.N.R., Naples, Italy DICMAPI – Università Federico II, Naples, Italy
h i g h l i g h t s The ignition process of propane mixtures is investigated at intermediate temperatures. The activation energy of the ignition process changes from intermediate to high temperatures. Such behavior strongly depends on the mixture dilution levels, stoichiometry and system pressure. The kinetics of the ignition process is numerically exploited. The competition between oxidative and recombination/pyrolytic routes promotes this behavior.
a r t i c l e
i n f o
Article history: Received 16 December 2014 Received in revised form 24 April 2015 Accepted 27 April 2015 Available online 5 May 2015 Keywords: Chemical kinetics Ignition delay times Numerical simulations MILD combustion Intermediate temperatures
a b s t r a c t The auto-ignition process of propane/oxygen mixtures has been widely studied in several facilities. Literature on shock tubes and tubular flow reactors has shown a change in the activation energy of the ignition process in the transition region from intermediate to high-temperature chemistry. Although fuel ignition chemistry has been widely exploited at low and high temperatures, further studies are required at intermediate temperatures. Based on previously published experimental results, this paper aimed to investigate the kinetics responsible for such a phenomenology, through a detailed kinetic analysis of the main pathways involved in the ignition process of propane mixtures. Simulations were performed over a wide range of temperatures (from low to high), changing the dilution levels of the mixture from ‘‘air’’ conditions to 97%, the pressure from 0.1 up to 3 MPa, and the mixture compositions from lean in fuel to rich in fuel. The analysis suggests that the phenomenology strongly depends on the mixture dilution levels and pressure. In particular, the differences in the ignition chemistry between low to intermediate temperatures and between intermediate to high temperatures is more evident for systems that are highly diluted and at low pressure. Such results are supported by data from the literature. This aspect explains why such a behavior is not reported for any experimental configurations and partially justifies the differences among the data obtained in several facilities, which is commonly addressed by the non-idealities of systems. Ó 2015 Elsevier B.V. All rights reserved.
1. Introduction Understanding the characteristic auto-ignition times and their relationship with operating conditions is vital for the successful design and dimensioning of combustion systems. This is particularly true for MILD combustion processes [1], in which highly diluted mixtures with inlet temperatures higher than that of fuel auto-ignition are used, thus giving rise to a combustion mode that
⇑ Corresponding author at: P.le Tecchio 80, 80125 Naples, Italy. Tel.: +39 0817683279; fax: +39 0812391709. E-mail address: [email protected] (P. Sabia). http://dx.doi.org/10.1016/j.cej.2015.04.143 1385-8947/Ó 2015 Elsevier B.V. All rights reserved.
is drastically different from that of traditional diffusive/deflagrative flames. In these applications, large mixture dilution levels are used to maintain working temperatures below critical values for the formation of pollutants species while high inlet temperatures are used to promote and sustain the oxidation process that occurs in homogeneous and nearly isothermal conditions [1–4]. In these conditions, the combustion regime is primarily controlled by a distributed auto-ignition process, which represents a key parameter in determining combustion behaviors [3–8]. However, high initial temperatures (above the auto-ignition value) and diluted conditions have significant effects on the chemical kinetics, affecting the auto-ignition chemistry [9–13]. Fuel nature, mixture
P. Sabia et al. / Chemical Engineering Journal 277 (2015) 324–333
composition, temperature and pressure represent the main controlling parameters of the ongoing chemical reactions in a combustion process. Many detailed studies have described the complex relationships of these parameters and their respective roles in determining combustion behavior. However, reports on these parameters under MILD conditions are scarce, and many chemical kinetics effects that are specific to these conditions must be further clarified. This information is essential for developing predictive, effective and robust numerical models for practical device design. Combined experimental and modeling efforts are required to elucidate the auto-ignition dependence on the operating parameters and to develop robust and predictable chemical kinetics models of MILD combustion processes. Many experimental and model aspects must be fixed to ensure the meaningfulness of obtained results. For instance, on the experimental side, the ignition process of hydrocarbons has been extensively studied over a wide range of operating conditions (pressure, temperature and mixture composition) and in many experimental facilities. Shock tubes [14–20], flow reactors [10,21–30] and rapid compression machines (RCM) [31,32] are commonly used to evaluate the ignition process. Although each of these methodologies has its merits, their utility is restricted to well-defined ranges of pressure, temperature and ignition times. Therefore, several experimental approaches have to be pursued in parallel. By combining data from different sources, the auto-ignition process of fuel/oxygen mixtures over a wide range of temperatures and pressures can be investigated. However, comparisons among data obtained through different experimental configurations or empirical/numerical methods must be made carefully. Interesting and even unexpected behaviors can emerge when the ignition process is experimentally studied under non-conventional conditions. Sabia et al. [10] numerically investigated methane oxidation under MILD conditions. They identified a phenomenology resembling ‘‘NTC’’ behavior of methane under MILD conditions at intermediate temperatures, and they identified the kinetics responsible for the behavior [9]. This phenomenology was previously experimentally identified by Huang et al. [20] in shock tubes for lean to rich methane/air mixtures but under different reference conditions, namely at high pressures (1.6–4.0 MPa) and at intermediate temperatures. Similarly, we previously [29] studied the propane auto-ignition process in a tubular flow reactor for highly diluted mixtures that were pre-heated to a wide range of temperatures. For mixtures characterized by fuel/oxygen ratios close to the stoichiometric conditions and diluted in N2 at 90%, at intermediate temperatures auto-ignition, delay times were almost independent of the inlet temperature. For high temperatures (Tin > 1100 K), the delay times linearly changed with the inlet temperature according to the Arrhenius plot. Previous reports of ignition delays from TFRs that obtained data for stoichiometric propane/air mixtures at several pressures have indicated a linear dependency on ignition delay temperature. Other recent studies [14–19] on propane auto-ignition in shock tubes have reported a sudden change in the activation energy when temperatures shift from intermediate to high at elevated pressures. In particular, Cadman et al. [14] measured propane ignition delay times at elevated pressures (0.5–4 MPa) for temperatures in the range of 850–1100 K for fuel-lean and stoichiometric mixtures in a shock tube. They compared the acquired ignition delay times with previous high temperature data provided by Burcart et al. [15,16], identifying a decrease in the ignition process activation energy in the range of 850 K < T < 1100 K. A sensitivity analysis demonstrated the importance of hydroperoxyl, propyl and methyl radicals at these intermediates temperatures. Similar
325
considerations were made by Herzler et al. [17]. They measured the ignition delay times of lean propane–air mixtures in the temperature range of 740–1300 K at pressures of 1 and 3 MPa in a shock tube facility. The experiments confirmed that the activation energy of the ignition delay time decreases at approximately 1050 K. Zhukov et al. [18] measured the ignition delay times behind reflected shock waves for lean propane–air mixtures in the temperature range of 800–1500 K and in the pressure range of 0.2– 50 MPa. They identified two different activation energies for the auto-ignition process that could be expressed as Arrhenius-dependent, at intermediate and high temperature ranges. They also updated a detailed kinetic mechanism based on their experimental data. Penyazkov et al. [19] measured the ignition delay times for lean, stoichiometric, and rich propane–air mixtures within the temperature range of 1000–1800 K at a pressure range of 0.2–2.0 MPa. They identified two different slopes for temperatures lower and higher than 1300 K. Healy et al. [31] performed several experiments in an RCM facility of mixtures composed of different proportions of methane/propane and methane/ethane/propane and investigated the effect of pressure on the ignition delay time. They compared their results with shock tube data obtained at high temperatures, observing a marked change in ‘‘activation energy’’ as a function of temperature in the transition region from intermediate- to high-temperature chemistry. In summary, propane reactivity is sensitive to changes from intermediate to high temperatures. This property is of interest regarding the ignition chemistry of hydrocarbons because propane exhibits thermo-chemical and combustion properties of larger hydrocarbon fuels more closely than either methane or ethane. Furthermore, propane oxidation is a sub-scheme of larger hydrocarbons, and thus understanding the ignition chemistry of such a reference fuel can provide useful information for a larger class of fuels. According to the previous considerations, the present work aimed to better define the oxidation chemistry of propane auto-ignition delay times by utilizing an extensive simulation of the chemical kinetics and comparing the numerical results with experimental data reported in the literature for several facilities and over a large range of operating conditions. The study focuses on the effects of the fuel/oxygen ratio, mixture dilution level and pressure at low, intermediate and high (750–1400 K) inlet temperature ranges. The objective is to identify possible chemical pathways responsible for the different auto-ignition trends observed across the various temperature ranges.
2. Critical analyses of test facilities Prior to analyzing the effects of mixture composition, dilution and pressure on the ignition data, a critical review on the reliability of ignition data is mandatory because non-idealities and perturbations may alter the data quality obtained from the reference facilities (shock tubes, tubular flow reactors and RCMs). A shock tube uses the compressive heating of a shock wave to rapidly increase the temperature and pressure of a premixed combustible mixture. Shock tubes are used to study the ignition process within short periods of times (on the order of tens of microseconds). Problems related to shock tubes have been widely discussed in the literature [32–35]. Recently, Davidson and Hanson [34] provided a clear and exhaustive analysis of shock tube performances. They identified the formation of a viscous boundary layer behind incident and reflected shock waves as a perturbing effect. This non-ideality causes a temperature and pressure increase in the core of the flow and attenuation of an incident shock wave,
326
P. Sabia et al. / Chemical Engineering Journal 277 (2015) 324–333
which significantly affects the apparent activation energy of the ignition process. Experiments are performed in argon, rather than nitrogen, to eliminate the possible influence of vibrational relaxation in the carrier gas and to eliminate or minimize shock wave bifurcation near the wall in the reflected shock front. Reflected shock temperatures are higher than those for nitrogen, influencing the ignition process timing history. Therefore, differences between combustion data in air and argon should be recognized and quantified. Furthermore the constant volume modeling approach is not valid because the energy released during the ignition process causes a rarefaction of the local pressure [35]. Such a condition is more stringent for highly concentrated fuel/oxygen mixtures for which the heat release is high. RCMs operate by compressing a homogeneous fuel/oxidizer mixture to moderate temperatures (up to Tmax 1200 K) and high pressures (Pmax 7 MPa) in a cylinder via the motion of a piston. The temperature and pressure can be sustained for longer than 10 ms. RCMs reproduce the processes that take place in devices such as spark engines and homogeneous charge compression ignition (HCCI) engines. Nevertheless, the time required to compress the test gas mixtures limits the minimum characteristic time of the investigation to 1 ms. One difficulty related to RCM operation is producing spatially homogenous conditions. Boundary layers, roll-up vortexes and heat loss to the walls during mixture compression can potentially change the temperature field in the cylinder and affect the measured ignition delay time [36–41]. Some strategies have been adopted [31,41] to avoid the formation of roll-up vortexes, but various non-idealities still remain. In a flow reactor (TFR), a pre-heated carrier gas flow rapidly mixes with a fuel flow, and the premixed charge enters a duct. The mixture chemically reacts as it flows downstream through the duct. Typical ignition times are on the order of milliseconds. The pre-heating temperatures are limited to electric heaters, and the pressures are moderate. To obtain useful kinetics data from flow reactor experiments, many of the secondary effects, such heat loss and species dispersion/diffusion, must be rendered negligible [22,23,27,28,42]. Furthermore, the mixing times must be shorter than the chemical kinetic times to insure true premixed conditions [42,43] at the entrance of the system. System non-idealities cannot be easily overcome, but the knowledge of their influence on data quality must be carefully addressed to interpret ignition data. The reference facilities may function properly under different operating conditions regarding pressure, inlet temperature and the range of ignition delay times. By combining data from these several facilities, it is possible to investigate the auto-ignition process of fuel/oxygen mixtures over a wide range of conditions.
3. Propane auto-ignition delay data Fig. 1 shows an Arrhenius plot of propane auto-ignition times (tign.) measured in several facilities that depicts stoichiometric conditions versus reactant inlet temperature (Tin). Although the figure does not show all available auto-ignition delay data on propane mixtures, it highlights the temperature dependence of various auto-ignition behaviors in several experimental configurations under different operating conditions. They are summarized in the upper part of Fig. 1. The ignition delay times are relative to a propane stoichiometric mixture in air conditions (approximately 76% in N2), except the data from Sabia et al. [29] where the mixture dilution level is 90% in N2, resembling MILD [1] conditions. Auto-ignition delay data in a RCM facility [31] show the typical NTC behavior of propane/air mixtures at low temperatures and high pressures. This behavior is not extensively discussed in the
Fig. 1. Literature auto-ignition delay data from several facilities at different operating conditions.
paper, but it was reported for completeness of the ignition phenomenologies related to propane mixtures. At higher inlet temperatures, data are generally obtained in TFR reactors. Fig. 1 shows data for a propane/oxygen stoichiometric mixture at several pressures, namely 0.1 MPa [22,26,27,29], 0.2 MPa [21], 0.9 MPa [28] and 1 MPa [25]. Under such operating conditions data show a linear trend with Tin, except the data by Holton et al. [27] and by Sabia et al. [29]. In such cases, the auto-ignition delay time curve shows two different slopes. In particular the data from [27] showed a linear trend of tign. with the inlet temperature for 1000/Tin > 1 in the Arrhenius plot, while for 1000/Tin < 1 the slope of the curve slightly diminishes. Sabia et al. [29] showed that for inlet temperatures lower than approximately 1100 K (1000/Tin = 0.9) tign. is almost independent on Tin, whereas for 1000/Tin < 0.9 it diminishes linearly with temperature. A similar behavior was reported in the literature for data obtained using shock tube facilities. In particular, Cadman et al. [14] identified a decrease in the ignition process activation energy in the range 850 < Tin < 1100 K compared to the high temperature auto-ignition values provided by Burcart et al. [15] at 0.5 MPa. Penyazkov et al. [19] also experimentally observed a change in the slope of the tign. vs Tin profiles at approximately 1300 K in the pressure range 1.2–2.0 MPa. These reports, obtained for stoichiometric propane mixtures in air at several pressures, suggest that the change in the ignition chemistry temperature also depends on pressure. Furthermore, results by Sabia et al. [29] indicated that highly diluted mixtures are more prone to show a change in slope between low and higher temperatures than undiluted (‘‘air’’) conditions at low pressures. These results, confirmed by those collected for fuel-lean mixtures in shock tube facilities [14,15,17] supports the idea that mixture dilution level also may play a role in the onset of different ignition regimes passing from low to intermediate and high temperatures. Sabia et al. [29] found experimentally that the change in the activation energy of propane ignition at intermediate temperatures depends also on the stoichiometry of the mixtures. In particular,
P. Sabia et al. / Chemical Engineering Journal 277 (2015) 324–333
they identified such a behavior from fuel-lean to rich mixtures in the range of 0.15 < C/O < 0.35 and mixture inlet temperatures from 975 to 1100 K. The variation of the activation energy of the autoignition process passing from intermediate to high temperatures for propane mixtures was mainly evidenced in studies realized in shock tubes under low-pressure conditions. In these operating conditions, measurements are most likely affected by the insurgence of non-ideal effects. Therefore, the auto-ignition dependence on the temperature was quite often attributed to these non-idealities [30,32,33,44]. On the other hand, the experiments by Holton et al. [27] and Sabia et al. [29] realized in tubular flow reactors have confirmed that propane mixtures exhibit a change of the activation energy of the auto-ignition process at intermediate temperatures where the non-linear effects can be neglected. A plausible alternative explanation of the observed behavior could be a change in the chemistry of propane mixtures leading to the ignition process. While the ‘‘NTC’’ behavior at low temperature has been widely investigated and the kinetics responsible of such a behavior completely identified, in literature there is not a thorough analysis of the phenomenologies observed at intermediate temperatures. The data from literature suggest that the change in the functional dependence of the auto-ignition on inlet temperatures from intermediate to high temperatures is complexly dependent on pressure (with mutual interactions), dilution level and mixture composition. The analysis of such a dependence thus requires a particular care and a thorough exploration of the variation of these parameters taking in consideration their possible interactions. 4. Numerical tools Simulations of the propane auto-ignition process in a plug-flow reactor were carried out using the PLUG application of ChemKin 3.7 [45] under adiabatic conditions. The absence of heat loss mechanisms from the reactor to the surroundings allows for the evaluation of only the kinetic aspects of the problem without considering complex interactions between heat exchange mechanisms and chemical reactions [10,11]. In agreement with previous studies [10,29], the mechanism by CRECK Modeling Group at Politecnico di Milano [46] was used to perform the simulations. It was validated in a wide range of operating conditions, from low temperatures, through the NTC behavior, up to high temperatures and pressures [47–49]. Before performing simulations, the kinetic mechanism was tested over a wide range of temperatures and pressures considering data from RCM, TFR and shock tube facilities. Fig. 2 shows a comparison among simulations and experimental data from the literature. Different criteria were applied to value the trustability of the kinetic mechanism on the basis of experimental
327
facility features and analytical procedure to value the ignition times. The RCM and shock tube auto-ignition delay times were simulated using the AURORA application of ChemKin 3.7 [45] and were considered as closed constant volume systems under adiabatic conditions. For such systems auto-ignition delay times were estimated on the basis of the temperature inflection point of the temporal temperature profile. The data from the tubular flow reactors were simulated using the PLUG application of ChemKin 3.7, considering no heat loss from the reactor to the surroundings, while the ignition criterion was based on a specified temperature increase beyond the initial temperature. In the case of TFR systems, the autoignition delay time was defined on the basis of the temperature profiles, considering as reference time the one corresponding to the abrupt temperature increase in air condition, while, in the case of diluted systems [29] a temperature increase beyond the inlet one of 10 K, according to previous studies [9,10,29]. In MILD combustion processes a criterion ad hoc has to be chosen [50]. Namely, under such operating conditions, the temperature increase during the oxidation process condition is relatively low because of high-diluted levels, and the temperature increase is not abrupt as in traditional flame conditions. Especially under ‘‘low reactive conditions’’, the temperature increase is very modest and the assumption of high temperature increase to define the ignition time would cause a net loss of information and the failure in identifying peculiar oxidation regimes. Considering the TFR system and the oxidation process features, we adopted a gradient of 10 K for the estimation of the auto-ignition delay time. The data reported by Beerer et al. [28] are relative to fuel-lean mixtures at 0.9 MPa, whereas the data reported by Sabia et al. [29] consider stoichiometric propane/oxygen mixtures diluted at 90% in N2 at environmental pressure. The ignition delay times by Penyazkov et al. [19] were obtained in a shock tube at 1.2– 2.0 MPa, whereas the data obtained by Cadman et al. [14] are relative to a lean propane/oxygen mixture (phi = 0.5) diluted in Ar at 0.5 MPa. The last data are different with respect to the one reported in Fig. 1 to value also the robustness of the kinetic scheme varying the mixture carbon/feed ratio. For low temperatures, the simulated data are relative to the RCM by Gallagher et al. [32] at 3 MPa. In such cases, the detailed mechanism provides a good approximation of the experimental data and is able to reproduce the NTC behavior. In such a region, the differences between numerical and experimental data are likely due to heat exchange effects that were not considered in the simulations. For TFR and shock tube data, the mechanism is able to reasonably reproduce both the auto-ignition delay time data and the slopes of curves. Some considerations apply for the data from shock tube facilities for both stoichiometric and fuel lean mixtures. In this context, it is worth noting that Sabia et al. [10,29] showed some discrepancies between their numerical and experimental data for fuel rich mixtures but the kinetic schemes well reproduced the dependence of the ignition delay time on the variation of this parameter. In general, under any condition, there is a reasonable agreement between simulations and experimental ignition delay times in the literature.
5. Numerical results
Fig. 2. Comparison among auto-ignition delay and simulation data reported in the literature.
The simulations were realized using the PLUG application of the ChemKin 3.7 software. In the next paragraphs, the auto-ignition delay times, the calculated for mixtures with different fuel/oxygen ratios, the mixture dilution levels and the pressures are reported.
328
P. Sabia et al. / Chemical Engineering Journal 277 (2015) 324–333
The analyses were performed primarily considering the operating conditions reported in Sabia et al. [29], namely the diluted conditions. The analyses are parametrically extended to ‘‘air’’ condition or to high pressures. This approach was chosen because under diluted conditions the kinetics is relatively slower with respect to ‘‘air’’ conditions and the different pathways involved in the auto-ignition process are easier discernable. This allows understanding the effects of the investigated parameters on the auto-ignition chemistry in a clearer way. 5.1. Effect of fuel/oxygen ratio Fig. 3 shows the auto-ignition delay times in the Arrhenius plot computed for fuel-rich, stoichiometric and fuel-lean conditions identified using C/O = 0.6, 0.3 and 0.05, respectively, at inlet temperatures from 800 to 1400 K at atmospheric pressure. The mixtures are diluted at 90% in N2. The figure shows that for low inlet temperatures, the auto-ignition delay times decrease from fuel-lean to fuel-rich mixtures. At temperatures below 1125 K (1000/Tin = 0.88), for the fuel-lean mixture, the auto-ignition delay values exhibit a linear dependence on temperature in the Arrhenius diagram plot. For the fuel stoichiometric and fuel-rich mixtures, there is a change in the slope of the auto-ignition times curves at approximately Tin = 950 K. Then, as for 1000/Tin = 0.8, the slopes of the curves change again, and for high inlet temperatures, fuel-lean mixtures correspond to lower ignition times. For high temperatures, the fuel-lean mixture exhibits a second change in slope, and thus the auto-ignition data shifts toward the stoichiometric curves. This behavior was experimentally determined by Qin et al. [51].
5.2. Effect of dilution levels Fig. 4 reports the auto-ignition time profiles in the Arrhenius plane for increasing dilutions of the stoichiometric mixtures with nitrogen, corresponding to standard ‘‘air’’ conditions (approximately 76%) to 97% nitrogen. The pressure is atmospheric. The auto-ignition delay times increase from a dilution level of 76% (air) to 97%. A change in slope is observed for all dilution values. It is worth noting that it becomes more evident as the nitrogen concentration increases. For instance, under ‘‘air’’ conditions, the change in the activation energy of the ignition process is almost unperceivable. Furthermore, the ignition profile inflection point shifts towards lower inlet temperatures with increasing mixture dilutions. In ‘‘air’’ conditions, the ignition process is sustained by high temperature gradients, and a fast chemistry dominates the process evolution. In contrast, for diluted conditions, the reactivity of the system slows because lower adiabatic flame temperatures make the system more sensitive to the operating conditions and enhances the competition among several pathways [29], thus permitting the onset of phenomenologies that are generally hidden during conventional combustion processes. 5.3. Effect of pressure The same simulations were performed for various pressures. Fig. 5 shows the numerical results for a stoichiometric mixture diluted in N2 at 90% at pressures ranging from 0.1 MPa to 3.0 MPa. The temperature range was extended to include low temperatures under the RCM conditions. In agreement with data from the literature, auto-ignition delay times decrease with increasing pressure. The slope variation between intermediate to high temperatures is less evident as the system pressure is increased and is shifted to higher inlet temperatures. In addition, the typical low temperature NTC behavior is observed at 2.0 MPa and becomes evident at 3.0 MPa over the same temperature range identified by Gallagher et al. [32]. 6. Discussion
Fig. 3. Auto-ignition delay data for fuel-lean, stoichiometric and fuel-rich conditions.
Fig. 4. Auto-ignition delay data for stoichiometric mixtures at 0.1 MPa with different dilution levels.
The numerical results reported in the previous section highlight unique behaviors related to the auto-ignition delay times of propane mixtures. In particular, the Arrhenius diagram plot of the auto-ignition times for intermediate temperatures exhibited a change in slope, which was dependent on the fuel/oxygen feed ratio, the mixture dilution level and the pressure. These results are supported by literature experimental data obtained from several facilities and operating conditions (Fig. 1). It is of uttermost interest to identify in the critical chemical kinetics pathways that
Fig. 5. Auto-ignition delay data for a stoichiometric mixture at several pressures.
329
P. Sabia et al. / Chemical Engineering Journal 277 (2015) 324–333
lead to the observed phenomenology. To this aim, standard chemical kinetics numerical tools, such as flux diagrams, and sensitivity and rate of production analyses of main species, are commonly used. In the following subsections, these methods will be presented, referencing the cases discussed in the previous section. 6.1. Main reactions involved in propane auto-ignition In Sabia et al. [29], the results relative to a stoichiometric mixture diluted at 90% at three different inlet temperatures, namely low, intermediate and high temperatures, at atmospheric pressure were reported. A comprehensive flux diagram is reported in Fig. 6 in order to better visualize and follow the different kinetic pathways. The flux diagram is relative to the auto-ignition time, identified on the basis of criterion earlier reported. Tree different colors identify the kinetic pathways at low (light gray), intermediate (dark gray) and high (black) temperature. Continuous lines refer to the carbon species pathways, while the dashed ones to radical production. They feed the branching mechanisms identified in the dashed-line rectangular insets. Reaction by reaction, the position of species reported aside any continuous line identifies the relative importance of reactants involved in the consumption of the carbon species reported in round frames.
At low temperatures, propane is dehydrogenated to isopropyl and normal-propyl by OH and HO2 radicals. Such C3 radicals mainly undergo dehydrogenation reactions, reacting with molecular oxygen (reactions 1 and 2 in Table 1). Propyl radicals also decompose to CH3 and C2H4 via reaction 3. Methyl radicals are oxidized to CH3O and CH2O via reactions 4–6, leading to the production of OH and H radicals. The HO2 radicals produced by reaction 1 and 2 primarily react through reactions 5 and 7. Finally, reaction 8
Table 1 Primary reactions involved in the auto-ignition process of propane mixtures. #
Reaction
1 2 3 4 5 6 7 8 9 10 11 12
O2 þ n-C3 H7 ) HO2 þ C3 H6 O2 þ i-C3 H7 ) HO2 þ C3 H6 n-C3 H7 ¼ CH3 þ C2 H4 CH3 þ CH3 OO ¼ CH3 O þ CH3 O CH3 þ HO2 ¼ CH3 O þ OH CH3 O þ M ¼ CH2 O þ H HO2 þ HO2 ¼ H2 O2 þ O2 H2 O2 ðþMÞ ¼ OH þ OHðþMÞ i-C3 H7 ¼ C3 H6 þ H CH3 þ CH3 þ M ¼ C2 H6 þ M H þ O2 ¼ OH þ O H þ O2 þ M ¼ HO2 þ M
Fig. 6. Comprehensive flux diagram at environmental pressure.
330
P. Sabia et al. / Chemical Engineering Journal 277 (2015) 324–333
represents the branching reaction at low temperature promoting the formation of OH radicals. At intermediate temperatures, while isopropyl radicals continue to react with oxygen to form HO2 radicals (reaction 1), propyl radicals, in contrast with the low temperature kinetics, primarily decompose to methyl radicals and ethylene (reaction 3). i-C3H7 also decomposes via an equilibrium reaction (reaction 9 in Table 1) that reconverts propene to i-C3H7 consuming H radicals. The branching reactions are related to H2O2 formation and decomposition, but the formation of HO2 radicals is relatively limited with respect to the previous flux diagram because n-C3H7 reacts via reaction 3 while i-C3H7 is only partially dehydrogenated by O2. The production of radical species is sustained by the pathway:
The typical high temperature branching reaction (reaction 11) becomes the branching mechanism in this temperature range and significantly increases the reactivity of the system, leading to a higher slope of auto-ignition delay time curve slope in the Arrhenius plot diagram with respect to intermediate temperatures. 6.2. Reaction rate analyses
Reaction 10 competes with the oxidation pathways, partially slowed by the relative depletion of HO2 radicals, lowering the production of OH and H radicals. This effect, combined with the less pronounced production of HO2, leads to lower reactivity of the system and to a smaller change in the slope of the auto-ignition time in the Arrhenius plot diagram. For high inlet temperatures, numerical analyses suggest that propane is dehydrogenated by OH and H radicals to normal and isopropyl species, decomposing via reactions 3 and 9, respectively. The last reaction boosts the production of H radicals. Propane also thermally decomposes to CH3 and C2H5. Methyl radicals mainly recombine to ethane feeding the pathways C2 H6 ) C2 H5 ) C2 H4 ) C2 H3 . The methyl oxidation routes are relatively less intense due to recombination reaction strong activation and to the relatively depletion of HO2 radicals.
The description of the kinetics involved in propane auto-ignition as a function of the inlet temperature is illustrated in Fig. 6, where the rate of key reactions, identified using sensitivity analyses (provided in Supplementary material) and flux diagrams, is reported as a function of the inlet temperature. In particular, Fig. 6a refers to the branching reactions, whereas Fig. 6b to C1 species. Fig. 7a describes the trend of reactions 7, 8 and 11. Furthermore, reaction 12 is reported in Table 1 because, under several operating conditions (e.g., high pressure), it competes with reaction 11. Notably, reactions 7 and 8 control the branching mechanism up to 1200 K, whereas reaction 11 is the most important reaction for higher inlet temperatures. Reaction 12 is negligible for any considered temperature at atmospheric pressure. Fig. 7b shows the relative weight of C1 oxidation and the recombination reactions. For Tin < 900 K, the C1 oxidation reactions, represented by reactions 4 and 5, are dominants. At higher temperatures, the recombination pathway (reaction 10) becomes progressively more relevant, and for Tin > 1100 K, this pathway is the dominant one. The onset of the recombination pathway slows the mixture reactivity because it favors the storage of C and H radicals in the form of C2 species. This inhibits the oxidation channel from producing radical species in a temperature range where the main branching mechanism (reaction 7 + 9) is relatively slow. This mechanism delays the auto-ignition process. At higher temperatures, reaction 11 dominates the branching mechanism, providing many radicals, accelerating the system reactivity, and inducing a change in the slope of the auto-ignition delay time profile. Similar analyses have been realized for fuel-rich and fuel-lean conditions.
Fig. 7. Reaction rate analysis for a stoichiometric C3H8/O2 mixture diluted in 90% N2 at atmospheric pressure.
Fig. 8. Reaction rate analysis for a fuel-rich (C/O = 0.6) C3H8/O2 mixture diluted in N2 at 90% at atmospheric pressure.
CH3 ) CH3 O ) CH2 O ) HCO ) CO In particular, methyl radicals are oxidized to CH3O by HO2 radicals. For Tin > 1000 K, reaction 10 begins to play an important role and promotes recombination and pyrolytic reactions, described by this sequence:
C2 H6 ) C2 H5 ) C2 H4 ) C2 H3
P. Sabia et al. / Chemical Engineering Journal 277 (2015) 324–333
Fig. 8 shows the rate of production of key reactions versus the inlet temperature relative to a rich mixture with a fuel/oxygen ratio equal to 0.6. Fig. 8a also shows the analysis relative to the rate of the key branching reactions, and Fig. 8b shows the reaction relative to C1 species. In both cases, the auto-ignition delay times versus inlet temperature are reported. As shown in Fig. 8b, for low temperatures, reaction 5 converts methyl radicals to CH3O and OH, which feed the oxidation channel:
CH3 ) CH3 O ) CH2 O ) HCO ) CO At 1000 K, the rate of reaction 5 equals that of reaction 10, thus delaying auto-ignition, as shown in the auto-ignition curve. At even higher temperatures, the recombination channel represents the main pathways of C1 species. Fig. 8a shows that the main branching mechanism up to 1200 K corresponds to reaction 8 (H2O2 (+M) = OH + OH (+M)), whereas for Tin > 1200 K, reaction 11 (H + O2 = OH + O) enhances the system reactivity and changes the slope of the auto-ignition delay time curve. For Tin > 1000 K, the rate of reaction 8 slowly increases, and for Tin > 1200 K, the reaction rate decreases due to the lack of HO2 radicals as C3H7 radicals are mainly decomposed and they do not react with O2, which leads to the production of HO2 radicals. In addition, CH3 radicals recombine to C2H6, which diminishes the production of radicals from the oxidation pathways. Fig. 9a and b show the rates of key reactions for fuel-lean mixtures. In reactions of fuel-lean mixtures, as also described in Fig. 3, the auto-ignition delay data exhibits a slight change in slope from intermediate to high temperatures. At 1100 K, the rates of reactions 5 and 10 are equal (Fig. 9a) while reaction 11 proceeds faster than reaction 8 (Fig. 9b). The delaying effect induced by reaction 10 is damped by the promotion of reaction 11, resulting in a slight change in the auto-ignition delay time curve. Furthermore, the flux diagrams and sensitivity analyses for fuel-lean mixtures (C/O = 0.05) (not reported here) have revealed that the competition between oxidation and recombination channels is altered by the promotion of other oxidative routes represented by the following reactions:
Fig. 9. Reaction rate analysis for a fuel-lean (C/O = 0.05) C3H8/O2 mixture diluted in N2 at 90% at atmospheric pressure.
331
1. CH3 + OH = CH3OH 2. CH3 + OH = CH2(S) + OH. The onset of such reactions, promoted by the high amount of OH radicals boosted by reaction 11, makes the recombination channel less active, thus lowering its effect on the ignition delay times. Furthermore, for temperatures higher than 1300 K, the considered reaction rates diminish. Such an effect is also observed in the trend of the auto-ignition delay times. The fast conversion of H to OH radicals by reaction 11 diminishes the concentration of H radicals that promptly dehydrogenate propane to its radicals at high temperatures. The rate of production analysis has been extended to stoichiometric mixtures diluted at 95% in N2 at atmospheric pressure to assess the influence of such a parameter on the kinetics of controlling the auto-ignition times. Fig. 10a and b are relative to the branching reactions and the C1 species. The reaction rate of the recombination reaction becomes higher than the oxidation one (represented by reaction 5) for Tin = 950 K. Therefore, because dilution promotes the recombination channel with respect to the oxidation routes for lower temperatures, the change in slope of the curve of the auto-ignition delay time occurs at lower temperatures with increasing dilution levels. Fig. 10a shows that the high temperature branching of reaction 11 becomes faster than reaction 8 for Tin = 1150 K, widening the range of temperatures for which reaction 8 is the primary branching reaction. Similar analyses were performed for stoichiometric mixtures diluted at 90% in N2 at 3.0 MPa. Fig. 11 shows the main kinetic reactions involved at the auto-ignition time as a function of inlet temperature for a stoichiometric mixture diluted in nitrogen at 90% at 3.0 MPa. Fig. 11b shows that the reaction CH3 + CH3OO = CH3O + CH3O mainly consumes CH3 radicals while reaction 11 becomes faster than reaction 5 at Tin = 1200 K. Therefore, the role of the recombination channel is primarily suppressed by reaction 4. Flux diagrams and sensitivity analyses showed that high pressures alter
Fig. 10. Reaction rate analysis for a stoichiometric C3H8/O2 mixture diluted in N2 at 95% at atmospheric pressure.
332
P. Sabia et al. / Chemical Engineering Journal 277 (2015) 324–333
Fig. 11. Reaction rate analysis for a stoichiometric C3H8/O2 mixture diluted in N2 at 90% at 3.0 MPa.
the equilibrium of reaction CH3 + O2 = CH3OO, producing more methyl peroxide radicals that in turn, participate in reaction 4 and boost the oxidation channel route. At high pressures, the fall-off reaction 10 (CH3 + CH3 + M = C2H6 + M) becomes bimolecular such that the dependence on the mixture dilution level becomes negligible, in contrast to diluted and low pressure systems. Fig. 11a shows that only at Tin = 1400 K, the rate of production of reaction 11 becomes higher than that of reaction 8. In addition, pressure promotes the competing reaction 12 that feeds reactions 7 and 8. The promotion of such kinetic pathways is more evident at higher pressures. Such consideration implies that the change in slope of the auto-ignition delay time curve is less evident and shifts to higher inlet temperatures when the pressure is increased. Previous reports have discussed the role of C1 recombination and oxidation routes and the competition among branching mechanisms. In addition, the prevailing kinetic route is strongly dependent on the C/O feed ratio, pressure and dilution levels. These parameters influence the relative production and distribution of radical species and the C1 oxidation/recombination competition mechanisms previously described.
7. Conclusions The auto-ignition process of propane/oxygen mixtures was numerically investigated with respect to mixture composition (carbon/oxygen ratio), dilution levels and pressure. Available literature data acquired using TFRs, shock tubes and RCMs have revealed a marked change in the ‘‘activation energy’’ of the auto-ignition process as a function of temperature in the region of transition from intermediate- to high-temperature chemistry. Such a behavior was reported in shock tubes for propane mixtures under low pressures and from intermediate to high temperatures. These operating conditions are not ideal for shock tube facilities, thus this behavior was mainly justified on the basis of non-idealities occurring in shock tube facilities exercised under these operating conditions.
Sabia et al. [29] reported the same behavior for diluted propane mixtures in a tubular flow reactor. Holton et al. [27] showed some other evidences of the insurgence of this behavior in a tubular flow reactor. An extensive kinetic analysis of this behavior has not been reported in literature. Thus the paper was devoted to the identification of the chemical pathways responsible of the onset of this phenomenology. Numerical results showed that this behavior is dependent on the fuel/oxygen feed ratio, mixture dilution levels and pressure confirming results from literature experimental data acquired from several facilities and under many operating conditions. The changes in auto-ignition times due to inlet temperature from intermediate to high temperatures depend on the operative parameters in addition to mutual, complex interactions. Therefore, propane reactivity is altered corresponding to intermediate versus high temperature regions. To identify the possible kinetic effects responsible for this phenomenology, a thorough numerical simulation analysis was undertaken here. The results of this analysis, as reported in the present study, show that the ignition chemistry is very sensitive to the competition between C1 oxidation and recombination/pyrolytic routes. Pressure, mixture dilution levels and composition alter the competition and thus promote different pathways. These parameters affect the relative production and distribution of radical species and the competition between C1 oxidation/recombination mechanisms.
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.cej.2015.04.143.
References [1] A. Cavaliere, M. de Joannon, Prog. Energy Combust. Sci. 30 (4) (2004) 329–366. [2] M. de Joannon, A. Saponaro, A. Cavaliere, M. de Joannon, Proc. Combust. Inst. 28 (2) (2000) 1639–1646. [3] I.B. Özdemir, N. Peters, Exp. Fluids 30 (2001) 683–695. [4] G.G. Szegö, B.B. Dally, G.J. Nathan, Combust. Flame 156 (2009) 429–438. [5] E. Oldenhof, M.J. Tummers, E.H. van Veen, D.J.E.M. Roekaerts, Combust. Flame 158 (8) (2011) 1553–1563. [6] M. de Joannon, G. Sorrentino, A. Cavaliere, Combust. Flame 159 (5) (2012) 1832–1839. [7] T. Plessing, N. Peters, J.G. Wünning, Proc. Combust. Inst. 27 (2) (1998) 3197– 3204. [8] C. Duwig, B. Li, Z.S. Li, M. Aldén, Combust. Flame 159 (1) (2012) 306–316. [9] P. Sabia, M. de Joannon, A. Picarelli, A. Chinnici, R. Ragucci, Fuel 91 (1) (2012) 238–245. [10] P. Sabia, M. de Joannon, A. Picarelli, A. Chinnici, R. Ragucci, Combust. Flame 160 (1) (2013) 47–55. [11] M. de Joannon, A. Cavaliere, T. Faravelli, E. Ranzi, P. Sabia, A. Tregrossi, Proc. Combust. Inst. 30 (2) (2005) 2605–2612. [12] A. Parente, C. Galletti, L. Tognotti, Int. J. Hydrogen Energy 33 (2008) 7553– 7564. [13] J. Aminian, C. Galletti, S. Shahhosseini, L. Tognotti, Flow Turbul. Combust. 88 (2012) 597–623. [14] P. Cadman, O.T. Geraint, P. Bulter, Phys. Chem. Chem. Phys. 2 (2000) 5411– 5419. [15] A. Burcat, K. Scheller, A. Lifshitz, Combust. Flame 16 (1) (1971) 29–33. [16] A. Burcat, K. Lifshitz, K. Scheller, G.B. Skinner, Proc. Combust. Inst. (1971) 745. [17] J. Herzler, L. Jerig, P. Roth, Combust. Sci. Technol. 176 (10) (2004) 1627–1637. [18] V.P. Zhukov, V.A. Sechenov, A.Yu. Starikovskii, Kinet. Catal. 46 (3) (2005) 319– 327. [19] O.G. Penyazkov, K.A. Ragotner, A.J. Dean, B. Varatharajan, Proc. Combust. Inst. 30 (2) (2005) 1941–1947. [20] J. Huang, W. Bushe, Combust. Flame 144 (1–2) (2006) 74–88. [21] V.Ya. Basevich, A.A. Borisov, S.M. Frolov, K.Ya. Troshin, G.I. Skachkov, in: Proceedings of the Fifth International Colloquium on Pulsed and Continuous Detonation, Moscow, Russia, 2006. [22] W. Freeman, A. Lefebvre, Combust. Flame 58 (1984) 153–164. 25. [23] A. Lefebvre, W. Freeman, L. Cowell, 1986, NASA CR-1750. [24] E. Lezberg, NACA TN-4028. 27, 1957. [25] L. Cowell, A. Lefebvre, SAE Technical Paper 860068, 1986.
P. Sabia et al. / Chemical Engineering Journal 277 (2015) 324–333 [26] C.J. Chang, A. Thompson, R. Winship, Ignition delay of propane in air between 725–800 °C under isothermal conditions, in: 7th Symposium (International) on Combustion, Academic Press, New York, 29 (1959) 431–435. [27] M.M. Holton, G.S. Jackson, P. Gokulakrihnan, M.S. Klassen, R.J. Roby, J. Eng. Gas Turbines Power 132 (9) (2010). [28] J. Beerer, V.G. McDonell, Proc. Combust. Inst. 33 (2011) 301–307. [29] P. Sabia, M. de Joannon, M. Lubrano Lavadera, P. Giudicianni, R. Ragucci, Combust. Flame 161 (12) (2014) 3022–3030. [30] A. Schönborn, P. Sayad, A.A. Konnov, J. Klingmann, Combust. Flame 160 (6) (2013) 1033–1043. [31] D. Healy, H.J. Curran, S. Dooley, J.M. Simmie, D.M. Kalitan, E.L. Petersen, G. Bourque, Combust. Flame 155 (3) (2008) 451–461. [32] S.M. Gallagher, H.J. Curran, W.K. Metcalfe, D. Healy, J.M. Simmie, G. Bourque, Combust. Flame 153 (1) (2008) 316–333. [33] L. Petersen, R.K. Hanson, Shock Waves 10 (2001) 405–420. [34] D.F. Davidson, R.K. Hanson, Int. J. Chem. Kinet. 36 (9) (2004) 510–523. [35] M. Chaos, F.L. Dryer, Int. J. Chem. Kinet. 42 (3) (2010) 143–150. [36] R. Minetti, M. Carlier, M. Ribaucour, E. Therssen, L.R. Sochet, Combust. Flame 102 (3) (1995) 298–309. [37] G. Mittal, M. Chomier, Combust. Flame 161 (2014) 75–83. [38] J. Clarkson, J.F. Griffiths, J.P. Macnamara, B.J. Whiteaker, Combust. Flame 125 (3) (2001) 1162–1175. [39] C.J. Sung, H.J. Curran, Prog. Energy Combust. Sci. 44 (2014) 1–18.
[40] [41] [42] [43] [44] [45]
[46] [47] [48]
[49] [50] [51]
333
G. Mittal, C.J. Sung, Combust. Flame 145 (2006) 160–180. G. Mittal, A. Bhari, Combust. Flame 160 (2013) 2975–2981. H. Mohamed, B. Hmaeid, Int. J. Energy Environ. Eng. 3 (22) (2012). H. Wu, M. Ihme, Combust. Flame 161 (2014) 2317–2326. K.-Y. Lam, Z. Hong, D.F. Davidson, R.K. Hanson, Proc. Combust. Inst. 33 (2011) 251–258. R.J. Kee, J.A. Miller, M.E. Coltrin, J.F. Grcar, E. Meeks, H.K. Moffat, A.E. Lutz, G. Dixon-Lewis, M.D. Smooke, J. Warnatz, G.H. Evans, R.S. Larson, R.E. Mitchell, L.R. Petzold, W.C. Reynolds, M. Caracotsios, W.E. Stewart, P. Glarborg, C. Wang, O. Adigun, W.G. Houf, C.P. Chou, S.F. Miller, Chemkin Collection, Release 3.7.1, Reaction Design, San Diego, CA, 2003. E. Ranzi, A. Frassoldati, R. Grana, A. Cuoci, T. Faravelli, A.P. Kelley, C.K. Law, Prog. Energy Combust. Sci. 38 (4) (2012) 468–501. E. Ranzi, T. Faravelli, P. Gaffuri, G.C. Pennati, A. Sogaro, Combust. Sci. Technol. 100 (1–6) (1994) 299–330. E. Ranzi, C. Cavallotti, A. Cuoci, A. Frassoldati, M. Pelucchi, T. Faravelli, Combust. Flame (2015), http://dx.doi.org/10.1016/j.combustflame.2014.11.030 (article in press). E. Ranzi, A. Frassoldati, A. Stagni, M. Pelucchi, A. Cuoci, T. Faravelli, Int. J. Chem. Kinet. 46 (9) (2014) 512–542. M. de Joannon, A. Cavaliere, R. Donnarumma, R. Ragucci, Proc. Combust. Inst. 29 (1) (2002) 1139–1146. Z. Qin, H. Yang, W.C. Gardiner, Combust. Flame 124 (2001) 246–254.