- Email: [email protected]
Autoignition delay times of propane mixtures under MILD conditions at atmospheric pressure
Combustion and Flame 161 (2014) 3022–3030
Contents lists available at ScienceDirect
Combustion and Flame j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / c o m b u s t fl a m e
Autoignition delay times of propane mixtures under MILD conditions at atmospheric pressure Pino Sabia a,⇑, Mariarosaria de Joannon a, Marco Lubrano Lavadera b, Paola Giudicianni b, Raffaele Ragucci a a b
Istituto di Ricerche sulla Combustione – C.N.R., Naples, Italy DICMAPI – Università Federico II, Naples, Italy
a r t i c l e
i n f o
Article history: Received 10 January 2014 Received in revised form 25 March 2014 Accepted 2 June 2014 Available online 3 July 2014 Keywords: MILD combustion Tubular flow reactor Dynamic behaviors Ignition delay times Chemical kinetics
a b s t r a c t The aim of the present work was to obtain experimental reference data in controlled, simple systems collected under MILD combustion. The combustion processes evolving under such conditions show behaviors specific to unique ranges of operating conditions that are not predictable using the available kinetic mechanisms. Experimental tests were conducted in a tubular flow reactor for propane/oxygen mixtures diluted in nitrogen under MILD combustion conditions by varying the mixture composition (from fuel-lean to fuel-rich conditions) and the dilution level over a wide range of temperatures (850-1250 K) at atmospheric pressure. Several combustion regimes were identified as a function of these external parameters. Auto-ignition delay times were evaluated, and they showed different levels of dependence on the system inlet temperature for intermediate to high temperatures. Under MILD combustion conditions, numerical simulations based on several available kinetic models predicted results with weak correlations to the experimental data, especially under fuel-rich, highly diluted conditions. To identify the controlling reaction pathways that should be tuned to extend the validity of models to wider operating conditions, sensitivity and reaction flux analyses were used. However, this topic is outside of the scope of the present work. This study provides reproducible experimental data for a reference system under novel conditions. Ó 2014 The Combustion Institute. Published by Elsevier Inc. All rights reserved.
1. Introduction MILD (Moderate or Intense Low-oxygen Dilution) combustion [1] technology has been accepted worldwide as a new combustion method of efficient and eco-friendly energy production. Such a combustion mode is achieved by using highly pre-heated and diluted mixtures to promote a homogeneous oxidation process under nearly isothermal conditions. The pre-heating levels promote the auto-ignition of the mixture and sustain a homogeneous oxidation process. Such operating conditions prevent the formation of pollutants (e.g., soot, NOx) [2–4], which are produced under very high temperatures in conventional flames, while ensuring the full conversion of fuels to combustion products. ⇑ Corresponding author. Address: P.le Tecchio 80, 80125 Naples, Italy. Fax: +39 0812391709. E-mail address: [email protected] (P. Sabia).
Coupling highly diluted mixtures with intense pre-heating alters the evolution of the combustion process with respect to traditional flames, thereby affecting the kinetics involved during fuel oxidation [5,6] (and thus the characteristic kinetic times) and the structure of the reactive region [7,8]. In the past [6–9], authors have focused on methane oxidation under MILD operative conditions in flow reactors. Their observations highlighted the onset of different kinetic routes promoted by high inlet temperatures and highly diluted mixtures with respect to conventional flames. Under such reference conditions, they characterized the ignition delay times [6] of methane in systems diluted in nitrogen in a tubular flow reactor. The auto-ignition time has practical relevance because it is one of the parameters essential for combustion design and dimensioning. Furthermore, an understanding of the kinetic routes established during auto-ignition is relevant for understanding the kinetic pathways of the oxidation process because reaction rates are relatively slow and very sensitive to operating conditions.
http://dx.doi.org/10.1016/j.combustflame.2014.06.006 0010-2180/Ó 2014 The Combustion Institute. Published by Elsevier Inc. All rights reserved.
P. Sabia et al. / Combustion and Flame 161 (2014) 3022–3030
Given this background, the aim of this work is to characterize oxidation regimes and the auto-ignition process of propane under diluted and pre-heated conditions. Propane reflects the thermochemical and combustion properties of larger hydrocarbons more accurately than methane or ethane. Because of these properties, it is a key fuel in experiments aimed at characterizing the oxidation kinetics of light to heavy hydrocarbons. The oxidation of propane has been extensively studied by numerous authors under a wide range of operating conditions (e.g., various pressures, temperatures and mixture compositions) in a wide range of experimental facilities (e.g., Shock Tubes, Rapid Compression Machines, tubular flow reactors). Apart open discussions regarding the sensitivity of ignition time measurements to experimental perturbations or non-idealities in reference systems [10–13], several auto-ignition time behaviors of propane have been identified as functions of temperature, pressure and mixture composition. Schultz and Shepherd [14] have provided an exhaustive review on the ignition delay times of propane in shock tube experiments. More recently, Gallagher et al. [15] reported an overview of autoignition data obtained in several facilities and under different operating conditions, along with a critical analysis of the reliability of data from Shock Tubes and RCMs. Experiments at low temperatures clearly showed the presence of Negative Temperature Coefficient (NTC) behavior [16–20]. Koert et al. [20] studied the oxidation of lean and diluted propane-oxygen mixtures in the negative temperature coefficient region of a plug flow reactor at pressures ranging from 10 to 15 atm, and they identified NTC behavior in the temperature range of 650–800 K. Recently, Gallagher et al. [15] studied the auto-ignition time in a rapid compression machine. They found NTC behavior in the temperature range of 750–850 K from 21 to 37 atm and at varying equivalence ratios from lean to rich. Other authors [20,21] have also discussed the specific oxidation kinetics responsible for NTC behavior at low temperatures. Studies at higher temperatures and pressures in Shock Tubes have reported changes in the temperature-dependence of the auto-ignition time for propane/oxygen mixtures passing from intermediate to high temperatures. In particular, Cadman et al. [22] measured propane ignition delay times at elevated pressures (5–40 bar) and temperatures (850–1100 K) for a fuel-lean (u = 0.5) mixture in a shock tube. These authors compared their ignition delay times with previous high temperature data [23], identifying a decrease in the activation energy of the ignition process at approximately 1000 K. Herzler et al. [24] used a shock tube to measure ignition delay times of lean (u = 0.5) propane–air mixtures in the temperature range of 740–1300 K at pressures of 10 and 30 bar. The obtained results were in very good agreement with the measurements of Cadman et al. [22]. The experiments confirmed that the activation energy of the ignition reactions decreases at approximately 1050 K. Zhukov et al. [25] measured ignition delay times behind reflected shock waves for lean (u = 0.5) propane–air mixtures in the temperature range of 800–1500 K and in the pressure range of 2–500 atm. They found two different activation energies for the auto-ignition process, expressed as an Arrhenius dependence, for intermediate and high temperature ranges. They also provided an updated, detailed kinetic mechanism to reflect their experimental data. Penyazkov et al. [26] measured the ignition delay times for lean, stoichiometric, and rich propane–air mixtures within a temperature range of 1000–1800 K and a pressure range of 2–20 atm. They provided an empirical correlation for ignition delay times by identifying two different slopes for temperatures lower and higher than 1300 K. Auto-ignition delay data have also been obtained in plug flow facilities at several pressures [27–35] in the intermediate
3023
temperature range. Studies have reported data with linear trends in Arrhenius plot diagrams. Beerer and McDonnell [34] performed propane auto-ignition delay time studies from 7 to 15 atm and from 785 to 935 K. They compared their results with shock tube data from the literature and suggested the establishment of two ignition regimes at intermediate to high temperatures. Recently Schönborn et al. [35] reported auto-ignition delay times for propane/air mixtures under conditions relevant to micro gas turbines for temperatures between 800 and 900 K at several pressures (0.4–0.6 MPa). Under such operating conditions, they reported a linear trend in auto-ignition times with temperature. They also provided an empirical correlation for auto-ignition delays under these operating conditions. While the propane auto-ignition process has been widely characterized under ‘‘air’’ conditions, there have been no experimental studies using MILD operating conditions at atmospheric pressures. The present work addresses propane combustion regimes and characteristic auto-ignition delay data under diluted and highly pre-heated conditions. Tests were conducted in a tubular flow reactor using propane/oxygen mixtures highly diluted in nitrogen at atmospheric pressure. Experimental data were obtained at intermediate to high temperatures. During the experimental tests, variations in the slope of the auto-ignition data versus the inlet temperature in the Arrhenius diagram were detected for mixture compositions close to stoichiometric conditions at intermediate to high temperatures. Similar observations were reported at high pressures in recent papers. This behavior was also detected at atmospheric pressure when mixtures were diluted and strongly pre-heated. This phenomenon has not been observed for ultra-lean fuel and rich mixtures, suggesting that the establishment of different kinetic routes that control auto-ignition chemistry is dependent on mixture compositions. 2. Experimental and numerical tools Experimental tests were carried out in a laboratory-scale stainless steel Tubular Flow Reactor (TFR) 140 cm in length with a 1 cm inner diameter [6,36]. A sketch of the experimental facility and a more accurate description of the reactor are included in the supplementary material. The main flow, composed of oxygen and diluent, is pre-heated to the desired temperature by means of electric fiber heaters. The pre-heated flow then mixes with the fuel flow by means of a ‘‘jets in cross flow’’ configuration [36]. The pre-mixed charge flows through the reactor where the auto-ignition and oxidation processes take place. Axially placed thermocouples (type N) permitted the recording of axial temperature profiles, which are used to evaluate oxidation regimes occurring during mixture and to measure ignition delay times (t). In particular, t is defined on the basis of axial steady temperature profiles. Once oxidation reactions occur, the temperature profile stabilizes after a transient period. The ignition time is then calculated: the axial distance over which a temperature increase of 10 K [37] is measured is divided by the flow velocity (defined as v = flow rate/reactor cross area). The experimental tests were carried out with carbon/oxygen (C/ O) ratios from 0 to 1 and inlet temperatures (Tin) from 850 to 1250 K. The mixture was diluted in nitrogen from 90% to 97%. Numerical simulations of the evolution of the propane oxidation process under diluted and highly preheated conditions in a tubular flow reactor were carried out using the PLUG module of ChemKin 3.7 [38]. The reactor overall heat transfer coefficient was calculated using the thermal resistances in series concept [39] on the basis of system boundary conditions, namely the oven
3024
P. Sabia et al. / Combustion and Flame 161 (2014) 3022–3030
temperature, in which the TFR is inserted to minimize heat exchange to the surroundings, and the inlet flow temperature and velocity, considering a non reactive case. Heat transfer coefficients was calculated by means of empirical correlations for flow in ducts [39]. The calculated value is 2.4 103cal/cm2 s K. Several detailed kinetic mechanisms were initially tested to determine their reliability for predicting experimental data. The simulations were then extended using a kinetic mechanism to highlight crucial aspects of auto-ignition delay times as a function of the mixture temperature at different dilution levels and mixture compositions. 3. Analysis of measurement reliability Before presenting the experimental results, discussions related to perturbations or effects of non-idealities on measurements are mandatory. In TFR systems, mixing times, thermocouple perturbations of fluid flow dynamics, and wall heterogeneous reactions are crucial. In flow systems, the mixing configuration should only allow for characteristic times shorter than the minimum ignition time when analyzing the oxidation process of premixed charges. The mixing section of the TFR was designed under this constraint, and the optimization procedure was performed following the previous literature, CFD calculations and experimental tests [36]. As previously stated, the mixing configuration uses a cross flow system. It allows for a characteristic mixing time shorter than 104 s (characteristic auto-ignition times were longer than 103 s). Furthermore, the mixing section was equipped with a thermocouple to monitor the temperature during flow mixing. When a temperature increase was detected by this thermocouple, indicating the onset of ignition/oxidation reactions during the mixing process, the experimental conditions were not considered. In such cases, the flow velocity was increased to decrease the mixing time, thereby increasing the local flow strain rate (and increasing the mixing efficiency and kinetic delay times) and shifting the ignition process further downstream in the TFR. The diameter of thermocouples was 1.5 mm. They were inserted in the reactor orthogonally to the flow. Radial temperatures profiles measurements for different inlet flow velocities showed that temperature profiles could be assumed to be flat starting from a radial position of 1 mm. To minimize the interference of thermocouples on fluid flow dynamics, the sensors were located at 2 mm from the wall. The invariance of the measured ignition delays for a range of flow velocities suggested that the thermocouples had no significant effects on the fluid flow. Thermocouples were located every 5 cm along the axial direction. The ignition data uncertainty due to thermocouple distance was approximately ±0.08% for a 30 m/s flow velocity, and ±0.035% for a 100 m/s flow velocity. Another issue to address in flow systems is the effect of heterogeneous reactions at walls. This aspect was thoroughly addressed in several ways: (1) Tests with mixtures of methane or propane with oxygen diluted in nitrogen were repeated after wall treatments in H2O or CO2 flows for hours. Furthermore, experimental tests were conducted in two tubes made from different types of stainless steel, namely AISI 316 and AISI 310S. Negligible differences in the ignition delay times were observed. (2) The literature suggests that catalytic effects can occur at low temperatures (thus, homogenous characteristic times are long) and long residence times (i.e., on the order of seconds). Because the tubular flow reactor typically operates at high temperatures and low residence times, it is reasonable to
infer that catalytic effects are not present or at least heterogeneous combustion times are significantly longer than homogeneous times. Indirect proofs of such a statement is given by the similar values of ignition delay times obtained at different flow rates and thus at different residence times. Heat exchange mechanisms with the surroundings can also affect the reliability of measurements. In the experimental tests, heat transfer did not significantly influence the ignition delay times for the considered operation conditions. By varying the inlet flux velocity, and thus the heat exchange coefficient, the change in the ignition time was lower than the uncertainty error introduced by thermocouple distance. Furthermore, several simulations were performed over a wide range of values for the global heat exchange coefficient. As in a previous study, no significant variations of the ignition delay time were identified for the temperatures considered in the experimental tests [6]. 4. Results An overview of several regimes that can be established in the ranges considered is provided by the analyses of axial temperature profiles collected at different inlet pre-heating temperatures, mixture compositions and dilution levels. In particular, several types of temperature profiles were observed and associated with characteristic system behaviors. Typical temperature profiles and the identification of combustion regimes are summarized in the supplementary material and in Sabia et al. [6]. Based on these classifications, a behavior map in a C/O-Tin plane was constructed and is shown in Fig. 1 for an N2 dilution of 90%. In particular, the map focuses on a temperature range between 850 K and 1150 K and C/O ratios from 0.025 to 1 for a 30 m/s inlet flow velocity. It is possible to distinguish several areas represented by different gray scale levels corresponding to unique axial profiles. At low temperatures (i.e., from 850 K to approximately 975 K), the mixtures with C/O values lower than the stoichiometric ratio did not ignite. The area is indicated as ‘‘no-combustion’’. The temperature profiles acquired for such conditions showed no
Fig. 1. Map of behaviors of C3H8/O2 mixtures diluted in N2 up to 90% and for v = 30 m/s.
P. Sabia et al. / Combustion and Flame 161 (2014) 3022–3030
temperature increases and remained equal to the inlet isothermal condition. For stoichiometric and fuel-rich mixtures, over the same temperature range, a ‘‘pyrolytic’’ behavior was observed. In such cases, the recorded temperature values were lower than the isothermal inlet profiles, suggesting that endothermic/pyrolytic reactions dominated. As the temperature increased, for C/O values in the neighborhood of the stoichiometric mixture (C/O = 0.3) the operating conditions led to ignition. For fuel-rich mixtures, the upper limit of the ‘‘ignition’’ region extends up to C/O = 0.35 and remains constant up to 1125 K, where it increases up to C/O = 0.5 at Tin = 1150 K. The lower limit of this area slightly extends towards lower C/O feed ratios as Tin increases. When the inlet temperature increased up to 1080 K, the ‘‘ignition’’ region extended to fuel ultra-lean conditions. Between the ‘‘ignition’’ and the ‘‘pyrolysis’’ regions, and between the ‘‘ignition’’ and the ‘‘no combustion’’ regions, the ‘‘low reactivity’’ behavior occurs. This refers to a condition in which the temperature increase (DT = T Tin) measured along the reactor axis is lower than 10 K with respect to the inlet temperature. In such a case, the ignition criterion is not satisfied despite sufficient fuel conversion. The pyrolytic line (dashed – dot) and the dynamic line (double dotted) are also shown in the map. All operating conditions above the pyrolytic line show a temperature decrease relative to the first thermocouples, indicating the onset of pyrolytic reactions. The operating conditions included in the ‘‘dynamic’’ line show oscillatory behaviors. The ‘‘dynamic’’ regime was identified when two unique temperature profiles were recorded for the same inlet conditions downstream of a steady ignition point. The last region in the map is relative to the ‘‘transient’’ behavior. For such a behavior, mixtures ignited temporarily, leading to an initial steady profile. Afterward, the temperature profile spontaneously shifted to a second and final state. Profiles relative to the first steady condition always show a DT > 10 K, while profiles related to the second steady condition may show a temperature increase greater than 10 K. When both the first and the second steady temperature profiles reach a DT > 10 K corresponding to operating conditions within their ignition regions on the map, two auto-ignition times were obtained as indicated by gray level intensities (transient I). In this case, the second steady state can exhibit a greater or lesser temperature increase than the first steady state. When the second steady profile did not satisfy the ‘‘ignition’’ criterion, the relative transient operating condition decreased into the low reactivity region (transient II). The ‘‘transient’’ region incorporates and shifts between the behaviors of several different kinetic controlling routes. The reactivity maps for other flow inlet velocities were also studied but are not reported here. They show that when the velocity is increased, the combustion regimes are still recognizable but are shifted to higher temperatures. Based on the temperature profiles used for the map in Fig. 1, the auto-ignition times (t) were evaluated for mixtures with 10 K temperature increases. The results are shown in Fig. 2 on a typical Arrhenius plot at 90% dilution and for several C/O ratios and flow velocities (i.e., from 30 to 70 m/s). For ultra-lean and lean conditions from 0 to 0.1, t shows a linear trend when plotted against temperature in the Arrhenius plot diagram. For C/O = 0.15 to the stoichiometric condition (C/O = 0.3), the auto-ignition delay time curves show two different slopes. In particular, for temperatures lower than approximately 1100 K, t is nearly independent from Tin. For 1000/Tin < 0.9, t diminishes linearly with temperature. For fuel-rich mixtures (C/O > 0.3), auto-ignition delay times vary linearly with temperature. Similarly, as with the reactivity
3025
maps, the ignition area for fuel-rich mixtures is narrow with respect to the inlet temperature parameter and mostly exhibits pyrolytic or transient behavior. The auto-ignition process is relatively slow with respect to lean/stoichiometric conditions; because of this, few data points were recorded at high temperatures. For transient conditions, the auto-ignition delay times are shown in Fig. 2. It is worth mentioning that t is nearly independent of the flow velocity. 5. Numerical analyses To understand the kinetics related to the auto-ignition process in MILD conditions, several numerical analyses were performed over a wide range of temperatures (from 800 K to 1400 K) at atmospheric pressure for a stoichiometric propane/oxygen mixture 90% diluted in nitrogen. Simulations were run using the PLUG application of the commercial software ChemKin 3.7.1 [38]. Several detailed kinetic mechanisms were evaluated for their ability to accurately predict the experimental data obtained in this work. The kinetic mechanisms are reported in Table 1, with the number of species and reactions included in the model along with the corresponding reference. Figure 3 shows comparisons between the experimental data and numerical results using the kinetic mechanisms of Table 1. Although most mechanisms predict changes in the slope of the auto-ignition delay time curve in the Arrhenius diagram in the intermediate temperature range, the predicted auto-ignition delay times for each kinetic scheme differ over all temperature ranges considered up to one order of magnitude. Additionally, four mechanisms predicted experimental autoignition delay times with the same order of magnitude. Among these, a few mechanisms predicted changes in the ignition behavior when passing from intermediate to high temperatures and were also able to predict, with good approximations, the experimental data collected in this work. The ‘‘Ranzi’’ mechanism, one of the more reliable predictors as shown in the figure and as reported in previous works [5,9], was selected as the kinetic model for additional numerical simulations in the auto-ignition process. The specific selection of the ‘‘Ranzi’’ mechanism does not alter the generality of this discussion: while elementary reactions in the other kinetic mechanisms have different kinetic parameters, their relative weights are similar, and the ‘‘main’’ and the ‘‘ignition controlling’’ reactions are the same. Figure 4 shows the auto-ignition delay times obtained for experimental and numerical simulations of fuel-lean (C/O = 0.05), fuel-rich (C/O = 0.6) and stoichiometric mixtures in the Arrhenius diagram. The dilution level was kept constant (90%). For the lean and stoichiometric conditions, the numerical simulation seems to agree with the experimental data, while for the fuel-rich condition, the experimental times are two or three times longer than the predicted data. In particular, the greater the C/O ratio (i.e., the richer the mixture), the greater the discrepancy between numerical simulations and experimental data. For inlet temperatures lower than approximately 1110 K, the numerical auto-ignition delay times were shortest for fuel-rich conditions and longest for fuel-lean mixtures. The stoichiometric mixture ignition time was between these values. As the inlet temperature increased above 1100 K, the fuel-rich mixture exhibited the slowest ignition time, whereas the fuel-lean mixture exhibited the most rapid time. The experimental data confirm that at approximately 1110 K, the lean mixture ignition values were lower than the stoichiometric ignition values. Figure 5 shows the experimental measurements and predicted numerical results of stoichiometric propane/oxygen mixtures
3026
P. Sabia et al. / Combustion and Flame 161 (2014) 3022–3030
Fig. 2. Auto-ignition delay times for C3H8/O2 mixtures from lean to rich conditions diluted in N2 to 90%.
Table 1 Detailed kinetic schemes used in simulations. Kinetic mechanism
Species
Reactions
Reference
‘‘Ranzi’’ GUI SAN DIEGO ‘‘Konnov’’ LLNL ‘‘Zhukov’’
82 293 50 127 155 209
1485 1593 244 1207 689 1260
[40] [41] [42] [43] [44] [45]
diluted in nitrogen to several levels (90%, 95% and 97%). There is good agreement between the experimental data and numerical predictions for the 90% dilution level. However, as the dilution level is increased to 95% and 97%, the correlation fails; specifically, the experimental data are longer than the predicted numerical results. Therefore, although the proposed kinetic mechanism is able to predict auto-ignition times for a few operating conditions, its ability to predict system behavior in MILD combustion processes is unreliable for fuel-rich mixtures and highly diluted mixtures.
Fig. 3. Comparison between experimental and numerical auto-ignition delay times for a stoichiometric C3H8/O2 mixture diluted in N2 to 90%.
6. Discussion Propane mixtures exhibited peculiar behaviors under MILD operating conditions.
P. Sabia et al. / Combustion and Flame 161 (2014) 3022–3030
Fig. 4. Experimental and numerical auto-ignition delay times for C3H8/O2 mixtures diluted in N2 to 90%.
3027
A second peculiar behavior was observed when the analysis of the auto-ignition times showed a change in the slope of ignition curves in the Arrhenius diagram at intermediate temperatures and C/O ratios of 0.15–0.3. This suggests that the competition among the different ignition regimes is very sensitive to mixture composition (Figs. 2 and 4). To understand the kinetics controlling the auto-ignition process for propane mixtures, numerical analyses were performed. Flux diagrams and sensitivity and rate of production analyses of the main species were carried out and reported for a stoichiometric mixture diluted up to 90% at three different inlet temperatures (Tin = 850 K, 1050 K and 1300 K, representing low, intermediate and high temperature conditions, respectively). Figure 6 shows the flux diagram at Tin = 850 K. The solid line represents the C-species (carbonaceous species), while dashed lines represent the reactions leading to the production of radicals. The reaction rate values are reported in square brackets and are normalized with respect to the scale indicated in the diagram. In the inset, the dashed-dotted line represents the main branching reactions. The main reactions involved in the ignition process of propane/ oxygen mixtures under MILD operating conditions are summarized in a table included in the supplementary material. The results suggest that at Tin = 850 K, propane is dehydrogenated to isopropyl and normal-propyl by OH and HO2 radicals. The resultant C3 radicals mainly undergo dehydrogenation reactions, reacting with molecular oxygen through reactions (1) and (2):
O2 þ nC3 H7 ) HO2 þ C3 H6
ð1Þ
O2 þ iC3 H7 ) HO2 þ C3 H6
ð2Þ
Normal-propyl radicals also decompose to CH3 and C2H4:
nC3 H7 ¼ CH3 þ C2 H4
ð3Þ
Methyl radicals are oxidized to CH3O and CH2O through the following reactions: Fig. 5. Experimental and numerical auto-ignition delay times for a stoichiometric C3H8/O2 mixture diluted in N2 to 90%, 95% and 97%.
CH3 þ CH3 OO ¼ CH3 O þ CH3 O
First, the reactivity map (Fig. 1) shows different oxidation regimes and dynamic behaviors over the analyzed ranges of temperature and mixture composition. Similarly, Sabia et al. [6] discussed these behaviors for methane. The authors affirmed that specific combustion regimes occur due to the interaction between the chemistry and heat-transfer mechanisms of the reactor. In conventional confined systems, the flame is stabilized by thermal feedback through wall-to-wall radiation, conduction in the tube wall and convection between the gas stream and the wall. In particular, axial heat conduction provides a thermal feedback mechanism (i.e., heat axial dispersion and conduction in the reactor wall) and can cause instabilities and multiple steady states. Under MILD operating conditions the heat release and temperature gradients are modest with respect to conventional flames; the pre-heating temperatures, which sustain the oxidation process, promote pyrolytic/recombination reactions; and the high dilution levels decrease the oxidation reaction rates. Both operating conditions lead to slower mixture reactivity with respect to conventional flames. Thus, the interaction between the heat exchanged to the surroundings and the heat produced by fuel oxidation is more dramatic. Such a heat balance leads to the establishment of several combustion regimes that are dependent on system parameters and the mixture composition. The slow transit between kinetic pathways, mainly due to the modest temperature gradients, is a critical point for the establishment of ‘‘transitional’’ and/or ‘‘dynamic’’ combustion regimes.
Fig. 6. Flux diagram for a stoichiometric C3H8/O2 mixture diluted in N2 to 90% with Tin = 850 K.
ð4Þ
3028
P. Sabia et al. / Combustion and Flame 161 (2014) 3022–3030
CH3 þ HO2 ¼ CH3 O þ OH
ð5Þ
CH3 O þ M ¼ CH2 O þ H
ð6Þ
Reaction 8 represents the branching reaction at low temperatures, promoting the formation of OH radicals. The flux diagram at Tin = 1050 K is shown in Fig. 7. At this inlet temperature, propane is dehydrogenated to its radicals i-C3H7 and n-C3H7. In this case, i-C3H7 can both react with oxygen (leading to the production of HO2 radicals) and decompose through an equilibrium reaction (9. iC3H7 = C3H6 + H). Decomposition leads to the production of H radicals, which mainly reconvert C3H6 to i-C3H7 at this Tin. However, n-C3H7 radicals mainly decompose to methyl radicals and ethylene (reaction 3). The branching reactions are still related to H2O2 formation and decomposition, but the formation of HO2 radicals is relatively limited with respect to Tin = 850 K. At this temperature, n-C3H7 does not react with oxygen while i-C3H7 decomposition (reaction 9) becomes relevant. In this case, the production of radical species is sustained by the pathway CH3 ) CH3 O ) CH2 O ) HCO ) CO. In particular, methyl radicals are oxidized to CH3O by HO2 radicals.
For Tin > 1000 K, the reaction CH3 + CH3 + M = C2H6 + M (10) plays an important role, promoting recombination and pyrolytic reactions, as described by the sequence C2 H6 ) C2 H5 ) C2 H4 ) C2 H3 . Reaction 10 competes with the oxidation pathways, the rates of which are reduced due to the depleted concentration of HO2 radicals. This results in decreased production of OH and H radicals. Such an effect, combined with the less pronounced production of HO2, leads to a relatively lower system reactivity and to changes in the slope of auto-ignition times in the Arrhenius plot diagram. Figure 8 shows the flux diagram at Tin = 1300 K. At this inlet temperature, propane is dehydrogenated by OH and H radicals to normal and iso-propyl species, which then decompose through reaction 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, thereby feeding the C2 H6 ) C2 H5 ) C2 H4 ) C2 H3 pathway. The methyl oxidation route is relatively less intense due to the strong activation energy of the recombination reaction and to the depletion of HO2 radicals. The typical high temperature branching reaction (11. H + O2 = OH + O) promotes auto-ignition at an inlet temperature of 1300 K. It significantly increases the reactivity of the system, leading to an increased auto-ignition delay time curve slope in the Arrhenius plot diagram compared with intermediate temperatures. The kinetics involved in propane auto-ignition as a function of Tin can also be deduced from Fig. 9. The rate of the key reaction is reported as a function of the inlet temperature along with t. In particular, Fig. 9a shows the auto-ignition relative to the H2/O2 sub-mechanism branching reactions, and Fig. 9b shows the autoignition relative to the C1 species.
Fig. 7. Flux diagram for a stoichiometric C3H8/O2 mixture diluted in N2 to 90% with Tin = 1050 K.
Fig. 8. Flux diagram for a stoichiometric C3H8/O2 mixture diluted in N2 to 90% with Tin = 1300 K.
leading to the production of OH and H radicals. HO2 radicals, produced by reactions 1 and 2, mainly react through reactions 5 and 7:
HO2 þ HO2 ¼ H2 O2 þ O2
ð7Þ
H2 O2 ðþMÞ ¼ OH þ OHðþMÞ:
ð8Þ
P. Sabia et al. / Combustion and Flame 161 (2014) 3022–3030
3029
7. Conclusion
Fig. 9. Rate of production analysis for key reactions for a stoichiometric C3H8/O2 mixture diluted in N2 to 90%.
Figure 9a describes the trend of reactions 7, 8 and 11. Furthermore, reaction 12 (H + O2 + M = HO2 + M) is included because it competes with reaction 11. It is worth noting that reactions 7 and 8 are the controlling branching mechanisms up to 1200 K. For higher inlet temperatures, reaction 11 is the most important mechanism. Reaction 12 does not act as a controlling step in this temperature range. Figure 9b shows the relative weight of C1 oxidation and recombination reactions. For Tin < 1100 K, the C1 oxidation reactions dominate through reactions 4 and 5. For Tin > 1100 K, the recombination reaction becomes the most important. The onset of reaction 10 slows down the mixture reactivity because it stores C and H radicals in the C2 species, thereby inhibiting the oxidation channel that produces radical species over a temperature range in which the main branching mechanism (reactions 7 and 8) is relatively weak. This mechanism increases autoignition delays. At higher temperatures, reaction 11 becomes the dominant branching mechanism, produces a large amount of radicals (thus accelerating the system reactivity), and induces a change in the slope of the ignition time curves. As a direct consequence of the competition among different kinetic routes, it is possible to infer that high levels of dilution lead to nearly isothermal conditions that, when associated with strong pre-heating, alter the relative weights of pyrolytic/oxidative routes, promoting the establishment of different regimes. When such competition becomes very sensitive to the operating conditions (i.e., the temperature and the carbon/oxygen mixture feed ratio), dynamic phenomena can emerge. Such behaviors are not recognizable in conventional systems because the high heat release, associated with standard deflagrated or diffused flames, enhances the shift among the different regimes, thereby promoting the establishment of kinetics typically observed at high temperatures. The numerical results here reported gives clear indications on the set of reactions to tune to improve the reliability of kinetic mechanisms to properly predict combustion features of small hydrocarbons for non-conventional conditions.
The exploitation of the oxidation process of propane mixtures under MILD conditions has led to the identification of different phenomena and combustion regimes. The analysis of experimental auto-ignition data clearly shows a marked change in the slopes of the Arrhenius plot for nearly stoichiometric mixtures. Additionally, dynamic phenomena related to the thermo-kinetic effects were identified for specific temperature ranges and C/O ratios. Using several kinetic models available in the literature, numerical predictions showed that only a few models were capable of only partially reproducing the observed experimental data. For fuel-rich mixtures and high dilution levels the numerical simulations do not accurately predict the experimental data. In this context, this study provided a database that can be used to update detailed kinetic schemes to extend their applicability to non-standard conditions. Numerical analyses were carried out to determine the kinetic pathways controlling the auto-ignition process to develop a refined model. The results show that the competition between C1 oxidation and recombination routes is responsible for the variation of the slope of the auto-ignition time in the Arrhenius diagram over an intermediate temperature range. Overall, high dilution levels decreased the rate of oxidation reactions, while high pre-heating temperatures promoted the recombination-pyrolytic routes, thereby increasing the competition between the two pathways. This analysis concludes that the oxidation and pyrolytic reactions should be properly tuned. Although this is beyond the scope of this work, our study has provided reproducible experimental data in a reference system under non-standard conditions. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.combustflame. 2014.06.006. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11]
[12] [13]
[14] [15] [16] [17] [18] [19] [20]
A. Cavaliere, M. de Joannon, Prog. Energy Combust. Sci. 30 (4) (2004) 329–366. J. Wünning, J. Wünning, Prog. Energy Combust. Sci. 23 (1) (1997) 81–94. R. Weber, J.P. Smart, Proc. Combust. Inst. 30 (2) (2005) 2623–2629. G. Szegö, B. Dally, G.J. Nathan, Combust. Flame 154 (1) (2008) 281–295. M. de Joannon, A. Cavaliere, T. Faravelli, E. Ranzi, P. Sabia, A. Tregrossi, Proc. Combust. Inst. 30 (2) (2005) 2605–2612. P. Sabia, M. de Joannon, A. Picarelli, R. Ragucci, Combust. Flame 160 (1) (2013) 47–55. M. de Joannon, A. Matarazzo, P. Sabia, A. Cavaliere, Proc. Combust. Inst. 31 (2) (2007) 3409–3416. M. de Joannon, P. Sabia, G. Cozzolino, G. Sorrentino, A. Cavaliere, Combust. Sci. Technol. 184 (7–8) (2012) 1207–1218. P. Sabia, M. de Joannon, A. Picarelli, A. Chinnici, R. Ragucci, Fuel 91 (1) (2012) 238–245. F.L. Dryer, M. Chaos, Combust. Flame 152 (1–2) (2008) 293–299. E. Petersen, M. Lamnaouer, J. de Vries, H. Curran, J. Simmie, M. Fikri, C. Schulz, G. Bourque, in: 26th International Symposium on Shock Waves, Springer, 2009, pp. 739–744. M. Chaos, F.L. Dryer, Int. J. Chem. Kin. 42 (2010) 143–150. C.J. Aul, E. Petersen, H. Curran, M. Fikri, C. Schulz, in: 23rd International Colloquium on the Dynamics of Explosions and Reactive Systems, July 24–29, 2011, Irvine CA.E. Schultz, J. Shepherd, California Institute of Technology, Pasadena, CA, 2000,
3030
P. Sabia et al. / Combustion and Flame 161 (2014) 3022–3030
[21] M. Cord, B. Husson, J.C.L. Huerta, O. Herbinet, P.-A. Glaude, R. Fournet, B. Sirjean, F. Battin-Leclerc, M. Ruiz-Lopez, Z. Wang, M. Xie, Z. Cheng, F. Qi, J. Phys. Chem. A 116 (50) (2012) 12214–12228. [22] P. Cadman, O.T. Geraint, P. Bulter, Phys. Chem. Chem. Phys. 2 (2000) 5411– 5419. [23] A. Burcat, K. Lifshitz, K. Scheller, G.B. Skinner, Proceedings Thirteenth Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, 1971. 745. [24] J. Herzler, L. Jerig, P. Roth, Combust. Sci. Technol. 176 (10) (2004). [25] V.P. Zhukov, V.A. Sechenov, A.Yu. Starikovskii, Kinet. Catal. 46 (3) (2005) 319– 327. [26] O.G. Penyazkov, K.A. Ragotner, A.J. Dean, B. Varatharajan, Proc. Combust. Inst. 30 (2) (2005) 1941–1947. [27] 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. [28] W. Freeman, A. Lefebvre, Combust. Flame 58 (1984) 153–162. [29] A. Lefebvre, W. Freeman, L. Cowell, (1986) NASA CR-1750. [30] E. Lezberg, (1957) NACA TN-4028. [31] L. Cowell, A. Lefebvre, (1986) SAE Technical Paper 860068. [32] C. Chang, A. Thompson, R. Winship, in: 7th Symposium (International) On Combustion, Academic Press, New York, 1959, pp. 431–435. [33] M.M. Holton, G.S. Jackson, P. Gokulakrihnan, M.S. Klassen, R.J. Roby, J. Eng. Gas Turbines Power 132 (9) (2010). [34] D.J. Beerer, V.G. McDonell, Proc. Combust. Inst. 33 (2011) 301–307. [35] A. Schönborn, P. Sayad, A.A. Konnov, J. Klingmann, Combust. Flame 130 (2013) 1033–1043.
[36] P. Sabia, Experimental and Numerical Studies of Mild Combustion processes in Model Reactors, PhD Thesis, 2006.. [37] M. de Joannon, A. Cavaliere, R. Donnarumma, R. Ragucci, Proc. Combust. Inst. 29 (1) (2002) 1139–1146. [38] F.M. Rupley, 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, Reaction Des., San Diego, CA, 2003. [39] F. Kreith, M.S. Bohn, Principles of Heat Transfer, sixth ed., Brooks/Cole, Pacific Grove, USA, 2001. [40] 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. . [42] ‘‘Chemical-Kinetic Mechanisms for Combustion Applications’’, San Diego Mechanism web page, Mechanical and Aerospace Engineering (Combustion Research), University of California at San Diego, . [43] A.A. Konnov, in: Twenty-Eighth Symposium (International) on Combustion, Edinburgh, Abstr. Symp. Pap., 2000, 317. [44] N.M. Marinov, W.J. Pitz, C.K. Westbrook, A.M. Vincitore, M.J. Castaldi, S.M. Senkan, Combust. Flame 114 (1998) 192–213, . [45] V.P. Zhukov, V.A. Sechenov, A. Yu. Starikovskii, Kinetics and Catalysis 46(3) (2005) 319–327, .
Contents lists available at ScienceDirect
Combustion and Flame j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / c o m b u s t fl a m e
Autoignition delay times of propane mixtures under MILD conditions at atmospheric pressure Pino Sabia a,⇑, Mariarosaria de Joannon a, Marco Lubrano Lavadera b, Paola Giudicianni b, Raffaele Ragucci a a b
Istituto di Ricerche sulla Combustione – C.N.R., Naples, Italy DICMAPI – Università Federico II, Naples, Italy
a r t i c l e
i n f o
Article history: Received 10 January 2014 Received in revised form 25 March 2014 Accepted 2 June 2014 Available online 3 July 2014 Keywords: MILD combustion Tubular flow reactor Dynamic behaviors Ignition delay times Chemical kinetics
a b s t r a c t The aim of the present work was to obtain experimental reference data in controlled, simple systems collected under MILD combustion. The combustion processes evolving under such conditions show behaviors specific to unique ranges of operating conditions that are not predictable using the available kinetic mechanisms. Experimental tests were conducted in a tubular flow reactor for propane/oxygen mixtures diluted in nitrogen under MILD combustion conditions by varying the mixture composition (from fuel-lean to fuel-rich conditions) and the dilution level over a wide range of temperatures (850-1250 K) at atmospheric pressure. Several combustion regimes were identified as a function of these external parameters. Auto-ignition delay times were evaluated, and they showed different levels of dependence on the system inlet temperature for intermediate to high temperatures. Under MILD combustion conditions, numerical simulations based on several available kinetic models predicted results with weak correlations to the experimental data, especially under fuel-rich, highly diluted conditions. To identify the controlling reaction pathways that should be tuned to extend the validity of models to wider operating conditions, sensitivity and reaction flux analyses were used. However, this topic is outside of the scope of the present work. This study provides reproducible experimental data for a reference system under novel conditions. Ó 2014 The Combustion Institute. Published by Elsevier Inc. All rights reserved.
1. Introduction MILD (Moderate or Intense Low-oxygen Dilution) combustion [1] technology has been accepted worldwide as a new combustion method of efficient and eco-friendly energy production. Such a combustion mode is achieved by using highly pre-heated and diluted mixtures to promote a homogeneous oxidation process under nearly isothermal conditions. The pre-heating levels promote the auto-ignition of the mixture and sustain a homogeneous oxidation process. Such operating conditions prevent the formation of pollutants (e.g., soot, NOx) [2–4], which are produced under very high temperatures in conventional flames, while ensuring the full conversion of fuels to combustion products. ⇑ Corresponding author. Address: P.le Tecchio 80, 80125 Naples, Italy. Fax: +39 0812391709. E-mail address: [email protected] (P. Sabia).
Coupling highly diluted mixtures with intense pre-heating alters the evolution of the combustion process with respect to traditional flames, thereby affecting the kinetics involved during fuel oxidation [5,6] (and thus the characteristic kinetic times) and the structure of the reactive region [7,8]. In the past [6–9], authors have focused on methane oxidation under MILD operative conditions in flow reactors. Their observations highlighted the onset of different kinetic routes promoted by high inlet temperatures and highly diluted mixtures with respect to conventional flames. Under such reference conditions, they characterized the ignition delay times [6] of methane in systems diluted in nitrogen in a tubular flow reactor. The auto-ignition time has practical relevance because it is one of the parameters essential for combustion design and dimensioning. Furthermore, an understanding of the kinetic routes established during auto-ignition is relevant for understanding the kinetic pathways of the oxidation process because reaction rates are relatively slow and very sensitive to operating conditions.
http://dx.doi.org/10.1016/j.combustflame.2014.06.006 0010-2180/Ó 2014 The Combustion Institute. Published by Elsevier Inc. All rights reserved.
P. Sabia et al. / Combustion and Flame 161 (2014) 3022–3030
Given this background, the aim of this work is to characterize oxidation regimes and the auto-ignition process of propane under diluted and pre-heated conditions. Propane reflects the thermochemical and combustion properties of larger hydrocarbons more accurately than methane or ethane. Because of these properties, it is a key fuel in experiments aimed at characterizing the oxidation kinetics of light to heavy hydrocarbons. The oxidation of propane has been extensively studied by numerous authors under a wide range of operating conditions (e.g., various pressures, temperatures and mixture compositions) in a wide range of experimental facilities (e.g., Shock Tubes, Rapid Compression Machines, tubular flow reactors). Apart open discussions regarding the sensitivity of ignition time measurements to experimental perturbations or non-idealities in reference systems [10–13], several auto-ignition time behaviors of propane have been identified as functions of temperature, pressure and mixture composition. Schultz and Shepherd [14] have provided an exhaustive review on the ignition delay times of propane in shock tube experiments. More recently, Gallagher et al. [15] reported an overview of autoignition data obtained in several facilities and under different operating conditions, along with a critical analysis of the reliability of data from Shock Tubes and RCMs. Experiments at low temperatures clearly showed the presence of Negative Temperature Coefficient (NTC) behavior [16–20]. Koert et al. [20] studied the oxidation of lean and diluted propane-oxygen mixtures in the negative temperature coefficient region of a plug flow reactor at pressures ranging from 10 to 15 atm, and they identified NTC behavior in the temperature range of 650–800 K. Recently, Gallagher et al. [15] studied the auto-ignition time in a rapid compression machine. They found NTC behavior in the temperature range of 750–850 K from 21 to 37 atm and at varying equivalence ratios from lean to rich. Other authors [20,21] have also discussed the specific oxidation kinetics responsible for NTC behavior at low temperatures. Studies at higher temperatures and pressures in Shock Tubes have reported changes in the temperature-dependence of the auto-ignition time for propane/oxygen mixtures passing from intermediate to high temperatures. In particular, Cadman et al. [22] measured propane ignition delay times at elevated pressures (5–40 bar) and temperatures (850–1100 K) for a fuel-lean (u = 0.5) mixture in a shock tube. These authors compared their ignition delay times with previous high temperature data [23], identifying a decrease in the activation energy of the ignition process at approximately 1000 K. Herzler et al. [24] used a shock tube to measure ignition delay times of lean (u = 0.5) propane–air mixtures in the temperature range of 740–1300 K at pressures of 10 and 30 bar. The obtained results were in very good agreement with the measurements of Cadman et al. [22]. The experiments confirmed that the activation energy of the ignition reactions decreases at approximately 1050 K. Zhukov et al. [25] measured ignition delay times behind reflected shock waves for lean (u = 0.5) propane–air mixtures in the temperature range of 800–1500 K and in the pressure range of 2–500 atm. They found two different activation energies for the auto-ignition process, expressed as an Arrhenius dependence, for intermediate and high temperature ranges. They also provided an updated, detailed kinetic mechanism to reflect their experimental data. Penyazkov et al. [26] measured the ignition delay times for lean, stoichiometric, and rich propane–air mixtures within a temperature range of 1000–1800 K and a pressure range of 2–20 atm. They provided an empirical correlation for ignition delay times by identifying two different slopes for temperatures lower and higher than 1300 K. Auto-ignition delay data have also been obtained in plug flow facilities at several pressures [27–35] in the intermediate
3023
temperature range. Studies have reported data with linear trends in Arrhenius plot diagrams. Beerer and McDonnell [34] performed propane auto-ignition delay time studies from 7 to 15 atm and from 785 to 935 K. They compared their results with shock tube data from the literature and suggested the establishment of two ignition regimes at intermediate to high temperatures. Recently Schönborn et al. [35] reported auto-ignition delay times for propane/air mixtures under conditions relevant to micro gas turbines for temperatures between 800 and 900 K at several pressures (0.4–0.6 MPa). Under such operating conditions, they reported a linear trend in auto-ignition times with temperature. They also provided an empirical correlation for auto-ignition delays under these operating conditions. While the propane auto-ignition process has been widely characterized under ‘‘air’’ conditions, there have been no experimental studies using MILD operating conditions at atmospheric pressures. The present work addresses propane combustion regimes and characteristic auto-ignition delay data under diluted and highly pre-heated conditions. Tests were conducted in a tubular flow reactor using propane/oxygen mixtures highly diluted in nitrogen at atmospheric pressure. Experimental data were obtained at intermediate to high temperatures. During the experimental tests, variations in the slope of the auto-ignition data versus the inlet temperature in the Arrhenius diagram were detected for mixture compositions close to stoichiometric conditions at intermediate to high temperatures. Similar observations were reported at high pressures in recent papers. This behavior was also detected at atmospheric pressure when mixtures were diluted and strongly pre-heated. This phenomenon has not been observed for ultra-lean fuel and rich mixtures, suggesting that the establishment of different kinetic routes that control auto-ignition chemistry is dependent on mixture compositions. 2. Experimental and numerical tools Experimental tests were carried out in a laboratory-scale stainless steel Tubular Flow Reactor (TFR) 140 cm in length with a 1 cm inner diameter [6,36]. A sketch of the experimental facility and a more accurate description of the reactor are included in the supplementary material. The main flow, composed of oxygen and diluent, is pre-heated to the desired temperature by means of electric fiber heaters. The pre-heated flow then mixes with the fuel flow by means of a ‘‘jets in cross flow’’ configuration [36]. The pre-mixed charge flows through the reactor where the auto-ignition and oxidation processes take place. Axially placed thermocouples (type N) permitted the recording of axial temperature profiles, which are used to evaluate oxidation regimes occurring during mixture and to measure ignition delay times (t). In particular, t is defined on the basis of axial steady temperature profiles. Once oxidation reactions occur, the temperature profile stabilizes after a transient period. The ignition time is then calculated: the axial distance over which a temperature increase of 10 K [37] is measured is divided by the flow velocity (defined as v = flow rate/reactor cross area). The experimental tests were carried out with carbon/oxygen (C/ O) ratios from 0 to 1 and inlet temperatures (Tin) from 850 to 1250 K. The mixture was diluted in nitrogen from 90% to 97%. Numerical simulations of the evolution of the propane oxidation process under diluted and highly preheated conditions in a tubular flow reactor were carried out using the PLUG module of ChemKin 3.7 [38]. The reactor overall heat transfer coefficient was calculated using the thermal resistances in series concept [39] on the basis of system boundary conditions, namely the oven
3024
P. Sabia et al. / Combustion and Flame 161 (2014) 3022–3030
temperature, in which the TFR is inserted to minimize heat exchange to the surroundings, and the inlet flow temperature and velocity, considering a non reactive case. Heat transfer coefficients was calculated by means of empirical correlations for flow in ducts [39]. The calculated value is 2.4 103cal/cm2 s K. Several detailed kinetic mechanisms were initially tested to determine their reliability for predicting experimental data. The simulations were then extended using a kinetic mechanism to highlight crucial aspects of auto-ignition delay times as a function of the mixture temperature at different dilution levels and mixture compositions. 3. Analysis of measurement reliability Before presenting the experimental results, discussions related to perturbations or effects of non-idealities on measurements are mandatory. In TFR systems, mixing times, thermocouple perturbations of fluid flow dynamics, and wall heterogeneous reactions are crucial. In flow systems, the mixing configuration should only allow for characteristic times shorter than the minimum ignition time when analyzing the oxidation process of premixed charges. The mixing section of the TFR was designed under this constraint, and the optimization procedure was performed following the previous literature, CFD calculations and experimental tests [36]. As previously stated, the mixing configuration uses a cross flow system. It allows for a characteristic mixing time shorter than 104 s (characteristic auto-ignition times were longer than 103 s). Furthermore, the mixing section was equipped with a thermocouple to monitor the temperature during flow mixing. When a temperature increase was detected by this thermocouple, indicating the onset of ignition/oxidation reactions during the mixing process, the experimental conditions were not considered. In such cases, the flow velocity was increased to decrease the mixing time, thereby increasing the local flow strain rate (and increasing the mixing efficiency and kinetic delay times) and shifting the ignition process further downstream in the TFR. The diameter of thermocouples was 1.5 mm. They were inserted in the reactor orthogonally to the flow. Radial temperatures profiles measurements for different inlet flow velocities showed that temperature profiles could be assumed to be flat starting from a radial position of 1 mm. To minimize the interference of thermocouples on fluid flow dynamics, the sensors were located at 2 mm from the wall. The invariance of the measured ignition delays for a range of flow velocities suggested that the thermocouples had no significant effects on the fluid flow. Thermocouples were located every 5 cm along the axial direction. The ignition data uncertainty due to thermocouple distance was approximately ±0.08% for a 30 m/s flow velocity, and ±0.035% for a 100 m/s flow velocity. Another issue to address in flow systems is the effect of heterogeneous reactions at walls. This aspect was thoroughly addressed in several ways: (1) Tests with mixtures of methane or propane with oxygen diluted in nitrogen were repeated after wall treatments in H2O or CO2 flows for hours. Furthermore, experimental tests were conducted in two tubes made from different types of stainless steel, namely AISI 316 and AISI 310S. Negligible differences in the ignition delay times were observed. (2) The literature suggests that catalytic effects can occur at low temperatures (thus, homogenous characteristic times are long) and long residence times (i.e., on the order of seconds). Because the tubular flow reactor typically operates at high temperatures and low residence times, it is reasonable to
infer that catalytic effects are not present or at least heterogeneous combustion times are significantly longer than homogeneous times. Indirect proofs of such a statement is given by the similar values of ignition delay times obtained at different flow rates and thus at different residence times. Heat exchange mechanisms with the surroundings can also affect the reliability of measurements. In the experimental tests, heat transfer did not significantly influence the ignition delay times for the considered operation conditions. By varying the inlet flux velocity, and thus the heat exchange coefficient, the change in the ignition time was lower than the uncertainty error introduced by thermocouple distance. Furthermore, several simulations were performed over a wide range of values for the global heat exchange coefficient. As in a previous study, no significant variations of the ignition delay time were identified for the temperatures considered in the experimental tests [6]. 4. Results An overview of several regimes that can be established in the ranges considered is provided by the analyses of axial temperature profiles collected at different inlet pre-heating temperatures, mixture compositions and dilution levels. In particular, several types of temperature profiles were observed and associated with characteristic system behaviors. Typical temperature profiles and the identification of combustion regimes are summarized in the supplementary material and in Sabia et al. [6]. Based on these classifications, a behavior map in a C/O-Tin plane was constructed and is shown in Fig. 1 for an N2 dilution of 90%. In particular, the map focuses on a temperature range between 850 K and 1150 K and C/O ratios from 0.025 to 1 for a 30 m/s inlet flow velocity. It is possible to distinguish several areas represented by different gray scale levels corresponding to unique axial profiles. At low temperatures (i.e., from 850 K to approximately 975 K), the mixtures with C/O values lower than the stoichiometric ratio did not ignite. The area is indicated as ‘‘no-combustion’’. The temperature profiles acquired for such conditions showed no
Fig. 1. Map of behaviors of C3H8/O2 mixtures diluted in N2 up to 90% and for v = 30 m/s.
P. Sabia et al. / Combustion and Flame 161 (2014) 3022–3030
temperature increases and remained equal to the inlet isothermal condition. For stoichiometric and fuel-rich mixtures, over the same temperature range, a ‘‘pyrolytic’’ behavior was observed. In such cases, the recorded temperature values were lower than the isothermal inlet profiles, suggesting that endothermic/pyrolytic reactions dominated. As the temperature increased, for C/O values in the neighborhood of the stoichiometric mixture (C/O = 0.3) the operating conditions led to ignition. For fuel-rich mixtures, the upper limit of the ‘‘ignition’’ region extends up to C/O = 0.35 and remains constant up to 1125 K, where it increases up to C/O = 0.5 at Tin = 1150 K. The lower limit of this area slightly extends towards lower C/O feed ratios as Tin increases. When the inlet temperature increased up to 1080 K, the ‘‘ignition’’ region extended to fuel ultra-lean conditions. Between the ‘‘ignition’’ and the ‘‘pyrolysis’’ regions, and between the ‘‘ignition’’ and the ‘‘no combustion’’ regions, the ‘‘low reactivity’’ behavior occurs. This refers to a condition in which the temperature increase (DT = T Tin) measured along the reactor axis is lower than 10 K with respect to the inlet temperature. In such a case, the ignition criterion is not satisfied despite sufficient fuel conversion. The pyrolytic line (dashed – dot) and the dynamic line (double dotted) are also shown in the map. All operating conditions above the pyrolytic line show a temperature decrease relative to the first thermocouples, indicating the onset of pyrolytic reactions. The operating conditions included in the ‘‘dynamic’’ line show oscillatory behaviors. The ‘‘dynamic’’ regime was identified when two unique temperature profiles were recorded for the same inlet conditions downstream of a steady ignition point. The last region in the map is relative to the ‘‘transient’’ behavior. For such a behavior, mixtures ignited temporarily, leading to an initial steady profile. Afterward, the temperature profile spontaneously shifted to a second and final state. Profiles relative to the first steady condition always show a DT > 10 K, while profiles related to the second steady condition may show a temperature increase greater than 10 K. When both the first and the second steady temperature profiles reach a DT > 10 K corresponding to operating conditions within their ignition regions on the map, two auto-ignition times were obtained as indicated by gray level intensities (transient I). In this case, the second steady state can exhibit a greater or lesser temperature increase than the first steady state. When the second steady profile did not satisfy the ‘‘ignition’’ criterion, the relative transient operating condition decreased into the low reactivity region (transient II). The ‘‘transient’’ region incorporates and shifts between the behaviors of several different kinetic controlling routes. The reactivity maps for other flow inlet velocities were also studied but are not reported here. They show that when the velocity is increased, the combustion regimes are still recognizable but are shifted to higher temperatures. Based on the temperature profiles used for the map in Fig. 1, the auto-ignition times (t) were evaluated for mixtures with 10 K temperature increases. The results are shown in Fig. 2 on a typical Arrhenius plot at 90% dilution and for several C/O ratios and flow velocities (i.e., from 30 to 70 m/s). For ultra-lean and lean conditions from 0 to 0.1, t shows a linear trend when plotted against temperature in the Arrhenius plot diagram. For C/O = 0.15 to the stoichiometric condition (C/O = 0.3), the auto-ignition delay time curves show two different slopes. In particular, for temperatures lower than approximately 1100 K, t is nearly independent from Tin. For 1000/Tin < 0.9, t diminishes linearly with temperature. For fuel-rich mixtures (C/O > 0.3), auto-ignition delay times vary linearly with temperature. Similarly, as with the reactivity
3025
maps, the ignition area for fuel-rich mixtures is narrow with respect to the inlet temperature parameter and mostly exhibits pyrolytic or transient behavior. The auto-ignition process is relatively slow with respect to lean/stoichiometric conditions; because of this, few data points were recorded at high temperatures. For transient conditions, the auto-ignition delay times are shown in Fig. 2. It is worth mentioning that t is nearly independent of the flow velocity. 5. Numerical analyses To understand the kinetics related to the auto-ignition process in MILD conditions, several numerical analyses were performed over a wide range of temperatures (from 800 K to 1400 K) at atmospheric pressure for a stoichiometric propane/oxygen mixture 90% diluted in nitrogen. Simulations were run using the PLUG application of the commercial software ChemKin 3.7.1 [38]. Several detailed kinetic mechanisms were evaluated for their ability to accurately predict the experimental data obtained in this work. The kinetic mechanisms are reported in Table 1, with the number of species and reactions included in the model along with the corresponding reference. Figure 3 shows comparisons between the experimental data and numerical results using the kinetic mechanisms of Table 1. Although most mechanisms predict changes in the slope of the auto-ignition delay time curve in the Arrhenius diagram in the intermediate temperature range, the predicted auto-ignition delay times for each kinetic scheme differ over all temperature ranges considered up to one order of magnitude. Additionally, four mechanisms predicted experimental autoignition delay times with the same order of magnitude. Among these, a few mechanisms predicted changes in the ignition behavior when passing from intermediate to high temperatures and were also able to predict, with good approximations, the experimental data collected in this work. The ‘‘Ranzi’’ mechanism, one of the more reliable predictors as shown in the figure and as reported in previous works [5,9], was selected as the kinetic model for additional numerical simulations in the auto-ignition process. The specific selection of the ‘‘Ranzi’’ mechanism does not alter the generality of this discussion: while elementary reactions in the other kinetic mechanisms have different kinetic parameters, their relative weights are similar, and the ‘‘main’’ and the ‘‘ignition controlling’’ reactions are the same. Figure 4 shows the auto-ignition delay times obtained for experimental and numerical simulations of fuel-lean (C/O = 0.05), fuel-rich (C/O = 0.6) and stoichiometric mixtures in the Arrhenius diagram. The dilution level was kept constant (90%). For the lean and stoichiometric conditions, the numerical simulation seems to agree with the experimental data, while for the fuel-rich condition, the experimental times are two or three times longer than the predicted data. In particular, the greater the C/O ratio (i.e., the richer the mixture), the greater the discrepancy between numerical simulations and experimental data. For inlet temperatures lower than approximately 1110 K, the numerical auto-ignition delay times were shortest for fuel-rich conditions and longest for fuel-lean mixtures. The stoichiometric mixture ignition time was between these values. As the inlet temperature increased above 1100 K, the fuel-rich mixture exhibited the slowest ignition time, whereas the fuel-lean mixture exhibited the most rapid time. The experimental data confirm that at approximately 1110 K, the lean mixture ignition values were lower than the stoichiometric ignition values. Figure 5 shows the experimental measurements and predicted numerical results of stoichiometric propane/oxygen mixtures
3026
P. Sabia et al. / Combustion and Flame 161 (2014) 3022–3030
Fig. 2. Auto-ignition delay times for C3H8/O2 mixtures from lean to rich conditions diluted in N2 to 90%.
Table 1 Detailed kinetic schemes used in simulations. Kinetic mechanism
Species
Reactions
Reference
‘‘Ranzi’’ GUI SAN DIEGO ‘‘Konnov’’ LLNL ‘‘Zhukov’’
82 293 50 127 155 209
1485 1593 244 1207 689 1260
[40] [41] [42] [43] [44] [45]
diluted in nitrogen to several levels (90%, 95% and 97%). There is good agreement between the experimental data and numerical predictions for the 90% dilution level. However, as the dilution level is increased to 95% and 97%, the correlation fails; specifically, the experimental data are longer than the predicted numerical results. Therefore, although the proposed kinetic mechanism is able to predict auto-ignition times for a few operating conditions, its ability to predict system behavior in MILD combustion processes is unreliable for fuel-rich mixtures and highly diluted mixtures.
Fig. 3. Comparison between experimental and numerical auto-ignition delay times for a stoichiometric C3H8/O2 mixture diluted in N2 to 90%.
6. Discussion Propane mixtures exhibited peculiar behaviors under MILD operating conditions.
P. Sabia et al. / Combustion and Flame 161 (2014) 3022–3030
Fig. 4. Experimental and numerical auto-ignition delay times for C3H8/O2 mixtures diluted in N2 to 90%.
3027
A second peculiar behavior was observed when the analysis of the auto-ignition times showed a change in the slope of ignition curves in the Arrhenius diagram at intermediate temperatures and C/O ratios of 0.15–0.3. This suggests that the competition among the different ignition regimes is very sensitive to mixture composition (Figs. 2 and 4). To understand the kinetics controlling the auto-ignition process for propane mixtures, numerical analyses were performed. Flux diagrams and sensitivity and rate of production analyses of the main species were carried out and reported for a stoichiometric mixture diluted up to 90% at three different inlet temperatures (Tin = 850 K, 1050 K and 1300 K, representing low, intermediate and high temperature conditions, respectively). Figure 6 shows the flux diagram at Tin = 850 K. The solid line represents the C-species (carbonaceous species), while dashed lines represent the reactions leading to the production of radicals. The reaction rate values are reported in square brackets and are normalized with respect to the scale indicated in the diagram. In the inset, the dashed-dotted line represents the main branching reactions. The main reactions involved in the ignition process of propane/ oxygen mixtures under MILD operating conditions are summarized in a table included in the supplementary material. The results suggest that at Tin = 850 K, propane is dehydrogenated to isopropyl and normal-propyl by OH and HO2 radicals. The resultant C3 radicals mainly undergo dehydrogenation reactions, reacting with molecular oxygen through reactions (1) and (2):
O2 þ nC3 H7 ) HO2 þ C3 H6
ð1Þ
O2 þ iC3 H7 ) HO2 þ C3 H6
ð2Þ
Normal-propyl radicals also decompose to CH3 and C2H4:
nC3 H7 ¼ CH3 þ C2 H4
ð3Þ
Methyl radicals are oxidized to CH3O and CH2O through the following reactions: Fig. 5. Experimental and numerical auto-ignition delay times for a stoichiometric C3H8/O2 mixture diluted in N2 to 90%, 95% and 97%.
CH3 þ CH3 OO ¼ CH3 O þ CH3 O
First, the reactivity map (Fig. 1) shows different oxidation regimes and dynamic behaviors over the analyzed ranges of temperature and mixture composition. Similarly, Sabia et al. [6] discussed these behaviors for methane. The authors affirmed that specific combustion regimes occur due to the interaction between the chemistry and heat-transfer mechanisms of the reactor. In conventional confined systems, the flame is stabilized by thermal feedback through wall-to-wall radiation, conduction in the tube wall and convection between the gas stream and the wall. In particular, axial heat conduction provides a thermal feedback mechanism (i.e., heat axial dispersion and conduction in the reactor wall) and can cause instabilities and multiple steady states. Under MILD operating conditions the heat release and temperature gradients are modest with respect to conventional flames; the pre-heating temperatures, which sustain the oxidation process, promote pyrolytic/recombination reactions; and the high dilution levels decrease the oxidation reaction rates. Both operating conditions lead to slower mixture reactivity with respect to conventional flames. Thus, the interaction between the heat exchanged to the surroundings and the heat produced by fuel oxidation is more dramatic. Such a heat balance leads to the establishment of several combustion regimes that are dependent on system parameters and the mixture composition. The slow transit between kinetic pathways, mainly due to the modest temperature gradients, is a critical point for the establishment of ‘‘transitional’’ and/or ‘‘dynamic’’ combustion regimes.
Fig. 6. Flux diagram for a stoichiometric C3H8/O2 mixture diluted in N2 to 90% with Tin = 850 K.
ð4Þ
3028
P. Sabia et al. / Combustion and Flame 161 (2014) 3022–3030
CH3 þ HO2 ¼ CH3 O þ OH
ð5Þ
CH3 O þ M ¼ CH2 O þ H
ð6Þ
Reaction 8 represents the branching reaction at low temperatures, promoting the formation of OH radicals. The flux diagram at Tin = 1050 K is shown in Fig. 7. At this inlet temperature, propane is dehydrogenated to its radicals i-C3H7 and n-C3H7. In this case, i-C3H7 can both react with oxygen (leading to the production of HO2 radicals) and decompose through an equilibrium reaction (9. iC3H7 = C3H6 + H). Decomposition leads to the production of H radicals, which mainly reconvert C3H6 to i-C3H7 at this Tin. However, n-C3H7 radicals mainly decompose to methyl radicals and ethylene (reaction 3). The branching reactions are still related to H2O2 formation and decomposition, but the formation of HO2 radicals is relatively limited with respect to Tin = 850 K. At this temperature, n-C3H7 does not react with oxygen while i-C3H7 decomposition (reaction 9) becomes relevant. In this case, the production of radical species is sustained by the pathway CH3 ) CH3 O ) CH2 O ) HCO ) CO. In particular, methyl radicals are oxidized to CH3O by HO2 radicals.
For Tin > 1000 K, the reaction CH3 + CH3 + M = C2H6 + M (10) plays an important role, promoting recombination and pyrolytic reactions, as described by the sequence C2 H6 ) C2 H5 ) C2 H4 ) C2 H3 . Reaction 10 competes with the oxidation pathways, the rates of which are reduced due to the depleted concentration of HO2 radicals. This results in decreased production of OH and H radicals. Such an effect, combined with the less pronounced production of HO2, leads to a relatively lower system reactivity and to changes in the slope of auto-ignition times in the Arrhenius plot diagram. Figure 8 shows the flux diagram at Tin = 1300 K. At this inlet temperature, propane is dehydrogenated by OH and H radicals to normal and iso-propyl species, which then decompose through reaction 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, thereby feeding the C2 H6 ) C2 H5 ) C2 H4 ) C2 H3 pathway. The methyl oxidation route is relatively less intense due to the strong activation energy of the recombination reaction and to the depletion of HO2 radicals. The typical high temperature branching reaction (11. H + O2 = OH + O) promotes auto-ignition at an inlet temperature of 1300 K. It significantly increases the reactivity of the system, leading to an increased auto-ignition delay time curve slope in the Arrhenius plot diagram compared with intermediate temperatures. The kinetics involved in propane auto-ignition as a function of Tin can also be deduced from Fig. 9. The rate of the key reaction is reported as a function of the inlet temperature along with t. In particular, Fig. 9a shows the auto-ignition relative to the H2/O2 sub-mechanism branching reactions, and Fig. 9b shows the autoignition relative to the C1 species.
Fig. 7. Flux diagram for a stoichiometric C3H8/O2 mixture diluted in N2 to 90% with Tin = 1050 K.
Fig. 8. Flux diagram for a stoichiometric C3H8/O2 mixture diluted in N2 to 90% with Tin = 1300 K.
leading to the production of OH and H radicals. HO2 radicals, produced by reactions 1 and 2, mainly react through reactions 5 and 7:
HO2 þ HO2 ¼ H2 O2 þ O2
ð7Þ
H2 O2 ðþMÞ ¼ OH þ OHðþMÞ:
ð8Þ
P. Sabia et al. / Combustion and Flame 161 (2014) 3022–3030
3029
7. Conclusion
Fig. 9. Rate of production analysis for key reactions for a stoichiometric C3H8/O2 mixture diluted in N2 to 90%.
Figure 9a describes the trend of reactions 7, 8 and 11. Furthermore, reaction 12 (H + O2 + M = HO2 + M) is included because it competes with reaction 11. It is worth noting that reactions 7 and 8 are the controlling branching mechanisms up to 1200 K. For higher inlet temperatures, reaction 11 is the most important mechanism. Reaction 12 does not act as a controlling step in this temperature range. Figure 9b shows the relative weight of C1 oxidation and recombination reactions. For Tin < 1100 K, the C1 oxidation reactions dominate through reactions 4 and 5. For Tin > 1100 K, the recombination reaction becomes the most important. The onset of reaction 10 slows down the mixture reactivity because it stores C and H radicals in the C2 species, thereby inhibiting the oxidation channel that produces radical species over a temperature range in which the main branching mechanism (reactions 7 and 8) is relatively weak. This mechanism increases autoignition delays. At higher temperatures, reaction 11 becomes the dominant branching mechanism, produces a large amount of radicals (thus accelerating the system reactivity), and induces a change in the slope of the ignition time curves. As a direct consequence of the competition among different kinetic routes, it is possible to infer that high levels of dilution lead to nearly isothermal conditions that, when associated with strong pre-heating, alter the relative weights of pyrolytic/oxidative routes, promoting the establishment of different regimes. When such competition becomes very sensitive to the operating conditions (i.e., the temperature and the carbon/oxygen mixture feed ratio), dynamic phenomena can emerge. Such behaviors are not recognizable in conventional systems because the high heat release, associated with standard deflagrated or diffused flames, enhances the shift among the different regimes, thereby promoting the establishment of kinetics typically observed at high temperatures. The numerical results here reported gives clear indications on the set of reactions to tune to improve the reliability of kinetic mechanisms to properly predict combustion features of small hydrocarbons for non-conventional conditions.
The exploitation of the oxidation process of propane mixtures under MILD conditions has led to the identification of different phenomena and combustion regimes. The analysis of experimental auto-ignition data clearly shows a marked change in the slopes of the Arrhenius plot for nearly stoichiometric mixtures. Additionally, dynamic phenomena related to the thermo-kinetic effects were identified for specific temperature ranges and C/O ratios. Using several kinetic models available in the literature, numerical predictions showed that only a few models were capable of only partially reproducing the observed experimental data. For fuel-rich mixtures and high dilution levels the numerical simulations do not accurately predict the experimental data. In this context, this study provided a database that can be used to update detailed kinetic schemes to extend their applicability to non-standard conditions. Numerical analyses were carried out to determine the kinetic pathways controlling the auto-ignition process to develop a refined model. The results show that the competition between C1 oxidation and recombination routes is responsible for the variation of the slope of the auto-ignition time in the Arrhenius diagram over an intermediate temperature range. Overall, high dilution levels decreased the rate of oxidation reactions, while high pre-heating temperatures promoted the recombination-pyrolytic routes, thereby increasing the competition between the two pathways. This analysis concludes that the oxidation and pyrolytic reactions should be properly tuned. Although this is beyond the scope of this work, our study has provided reproducible experimental data in a reference system under non-standard conditions. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.combustflame. 2014.06.006. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11]
[12] [13]
[14] [15] [16] [17] [18] [19] [20]
A. Cavaliere, M. de Joannon, Prog. Energy Combust. Sci. 30 (4) (2004) 329–366. J. Wünning, J. Wünning, Prog. Energy Combust. Sci. 23 (1) (1997) 81–94. R. Weber, J.P. Smart, Proc. Combust. Inst. 30 (2) (2005) 2623–2629. G. Szegö, B. Dally, G.J. Nathan, Combust. Flame 154 (1) (2008) 281–295. M. de Joannon, A. Cavaliere, T. Faravelli, E. Ranzi, P. Sabia, A. Tregrossi, Proc. Combust. Inst. 30 (2) (2005) 2605–2612. P. Sabia, M. de Joannon, A. Picarelli, R. Ragucci, Combust. Flame 160 (1) (2013) 47–55. M. de Joannon, A. Matarazzo, P. Sabia, A. Cavaliere, Proc. Combust. Inst. 31 (2) (2007) 3409–3416. M. de Joannon, P. Sabia, G. Cozzolino, G. Sorrentino, A. Cavaliere, Combust. Sci. Technol. 184 (7–8) (2012) 1207–1218. P. Sabia, M. de Joannon, A. Picarelli, A. Chinnici, R. Ragucci, Fuel 91 (1) (2012) 238–245. F.L. Dryer, M. Chaos, Combust. Flame 152 (1–2) (2008) 293–299. E. Petersen, M. Lamnaouer, J. de Vries, H. Curran, J. Simmie, M. Fikri, C. Schulz, G. Bourque, in: 26th International Symposium on Shock Waves, Springer, 2009, pp. 739–744. M. Chaos, F.L. Dryer, Int. J. Chem. Kin. 42 (2010) 143–150. C.J. Aul, E. Petersen, H. Curran, M. Fikri, C. Schulz, in: 23rd International Colloquium on the Dynamics of Explosions and Reactive Systems, July 24–29, 2011, Irvine CA.E. Schultz, J. Shepherd, California Institute of Technology, Pasadena, CA, 2000,
3030
P. Sabia et al. / Combustion and Flame 161 (2014) 3022–3030
[21] M. Cord, B. Husson, J.C.L. Huerta, O. Herbinet, P.-A. Glaude, R. Fournet, B. Sirjean, F. Battin-Leclerc, M. Ruiz-Lopez, Z. Wang, M. Xie, Z. Cheng, F. Qi, J. Phys. Chem. A 116 (50) (2012) 12214–12228. [22] P. Cadman, O.T. Geraint, P. Bulter, Phys. Chem. Chem. Phys. 2 (2000) 5411– 5419. [23] A. Burcat, K. Lifshitz, K. Scheller, G.B. Skinner, Proceedings Thirteenth Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, 1971. 745. [24] J. Herzler, L. Jerig, P. Roth, Combust. Sci. Technol. 176 (10) (2004). [25] V.P. Zhukov, V.A. Sechenov, A.Yu. Starikovskii, Kinet. Catal. 46 (3) (2005) 319– 327. [26] O.G. Penyazkov, K.A. Ragotner, A.J. Dean, B. Varatharajan, Proc. Combust. Inst. 30 (2) (2005) 1941–1947. [27] 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. [28] W. Freeman, A. Lefebvre, Combust. Flame 58 (1984) 153–162. [29] A. Lefebvre, W. Freeman, L. Cowell, (1986) NASA CR-1750. [30] E. Lezberg, (1957) NACA TN-4028. [31] L. Cowell, A. Lefebvre, (1986) SAE Technical Paper 860068. [32] C. Chang, A. Thompson, R. Winship, in: 7th Symposium (International) On Combustion, Academic Press, New York, 1959, pp. 431–435. [33] M.M. Holton, G.S. Jackson, P. Gokulakrihnan, M.S. Klassen, R.J. Roby, J. Eng. Gas Turbines Power 132 (9) (2010). [34] D.J. Beerer, V.G. McDonell, Proc. Combust. Inst. 33 (2011) 301–307. [35] A. Schönborn, P. Sayad, A.A. Konnov, J. Klingmann, Combust. Flame 130 (2013) 1033–1043.
[36] P. Sabia, Experimental and Numerical Studies of Mild Combustion processes in Model Reactors, PhD Thesis, 2006.