A 5-step reduced mechanism for combustion of CO/H2/H2O/CH4/CO2 mixtures with low hydrogen/methane and high H2O content

TL;DR: ZMN and NS acknowledge the funding through the Low Carbon Energy University Alliance Programme supported by Tsinghua University, China as mentioned in this paper, and also like to acknowledge the educational grant through the A.G. LeventisFoundation.

Abstract: ZMN and NS acknowledges the funding through the Low Carbon Energy University Alliance Programme supported by Tsinghua University, China. ZMN also likes to acknowledge the educational grant through the A.G. Leventis Foundation.

Summary

1. Introduction

• Recent developments in gas-turbine power generation include the use of low calorific value fuels.
• Generally, these mechanisms were developed systematically by introducing steady-state and/or partial equilibrium assumptions respectively for some species and reactions involved in a skeletal mechanism.
• The hydrogen content is low in the BFG as noted earlier and one may like to mix it with small amounts of H2, CH4 and H2O or other gases containing high fractions of these species in order to enhance the BFG combustion characteristics.
• To the best of their knowledge this is the first attempt to obtain a reduced mechanism for a multi-component fuel mixture with a good accuracy over a wide range of thermodynamic and thermo-chemical conditions.

2. Development of skeletal mechanism: Sensitivity analysis

• The chemical kinetics of CO/H2 mixture oxidation has been investigated by numerous studies in the past and a sustained interest on the combustion of Syngas in gas turbines for power generation has led to publication of a dedicated volume on this topic in the Combustion Science and Technology journal in 2008.
• Flame speed sensitivity analyses are conducted using the GRI [19], at high (20%) and zero water vapour content in the fuel mixture in order to (1) identify the most important reactions in each case and (2) to obtain a suitable skeletal mechanism for CO/H2/H2O mixtures.
• CO2 has large sensitivities for both dry 9 and wet mixtures and OH + CO = H+ CO2 remains as the most important reaction with sensitivity nearly five times larger than for the HO2 reaction for CO consumption.
• Thus, the effect of small CH4 amounts in the fuel mixture is adequately captured by the extra 9 reactions noted above, something which was neglected while developing reduced mechanism in a previous study [14].

3. Development of reduced chemistry

• By removing certain intermediate species from the detailed mechanism, the computational effort is reduced as the number of ODEs that must be solved is decreased.
• Intermediate species can be systematically identified and removed from the ODE system via two major sequential steps.
• Second, further reduction of the skeletal mechanism results in a reduced mechanism.
• For fast development of reduced chemistry, the interactive Computer Assisted Reduction Mechanism (CARM) algorithm [40, 43] was used for the automatic generation of reduced chemistry with the ability to produce source codes needed for computing the chemical sources.

4. Reduced mechanism

• The same would apply in cases where Ar is the inert.
• Also, for fine tuning of the reduced chemistry, the activation energy of reaction 2 in Table 3 was increased by 27.5%, a procedure similar to the correction factor employed by Boivin et al. [14] to correctly predict the ignition delay times.
• The steady-state relationships can be written as dCA dt = ψA(ss, ss )− gA(ss, ss)CA = 0, where ψA(ss, ss) and gA(ss, ss) are functions of species both in steady-state, denoted by ss, and non steady-state, denoted by ss.
• Since the current QSS species are not strongly coupled, the point iteration scheme is found to be sufficient for the present case.

5. Validation

• Both the skeletal and reduced mechanisms are validated over wide range of conditions shown in Table 4, by comparing laminar flame speeds, ignition delay times and the flame structure with experimental results and/or the computational results obtained using the GRI Mech 3.0 [19].
• The flame speeds are calculated using the PREMIX [45] code of the CHEMKIN package [46] including the thermal diffusion and multi-component formulation for the species’ diffusivities.
• In the cases where no experimental data are available, the skeletal and the reduced mechanisms are validated against the predictions of the GRI Mech 3.0 [19] and so readers are cautioned while interpreting this particular comparison.
• In calculating the ignition delay times with the reduced mechanism, the correction factor used in the study of Boivin et 18 al. [14] is employed.
• This correction factor was originally developed in [47] from an analysis of the autoignition eigenvalue under lean conditions.

5.1. Premixed flames

• Comparisons of computed flame speeds, sL, against available experimental data for the mixtures listed in Table 4 are presented in Figs. 1-10.
• The above comparisons show that overall both the skeletal and the reduced mechanism give good agreement with the experimental data and the computations with the GRI Mech 3.0 [19].
• The skeletal mechanism of [14] as implemented in this study, under-predicts the flame speeds for all equivalence ratios and the level of under-prediction increases with the H2O content in the fuel mixture.
• Again there is a good agreement with the full GRI Mech 3.0 [19] and it is somewhat improved in the high pressure case, compared to the predictions of the methane-containing fuel mixture in Fig. 11.

5.2. Autoignition

• Figure 21 compares the computed ignition delay times (with the correction factor in Eq. 3 applied) with the experimental results of Kalitan et al. [55] for CO/H2 mixtures over a range of conditions listed in Table 4.
• Overall, the agreement is very good for both low and high pressures and for the entire range of temperatures considered.
• Figure 22 compares ignition delay times computed for a CO2-diluted mixture to the measured values in [28] at different pressures.
• The reduced mechanism shows good agreement with the experimental data for the entire temperature range.
• As noted in [28] using sensitivity analysis, the most important reactions at the conditions tested were the chain-branching reactions and the three body recombination reaction H + O2 + CO2 = HO2 + CO2.

6. Speed up times

• Table 5 shows the time in seconds taken for each run for each of the conditions shown in Table 4.
• The flame speeds were calculated using the PREMIX code [45] with thermal diffusion and a multi-component formulation for the species’ diffusivities, in a 2.5 cm domain with adaptive grid.
• It is clear that both the skeletal and reduced mechanisms reduced the computational time significantly compared to the GRI Mech 3.0 [19], while maintaining the same level of accuracy.
• In particular for case 3 the skeletal mechanism is about 50 times faster and the reduced mechanism about 300 times faster.

7. Conclusions

• A 5-step reduced chemical kinetic mechanism involving 9 species for accurate prediction of the combustion characteristics of multi-species fuel mixtures of CO/H2/H2O/CH4/CO2, having low hydrogen/methane and high water vapour content is derived.
• These two mechanisms are tested for their ability to predict laminar flame speeds, flame structure and ignition delay times over a wide range of pressure, temperature and fuel mixture composition.
• It is also worth to note that these conditions are relevant for stationary gas-turbines for power generation.
• Furthermore, it is found that use of the reduced mechanism decreases the computational time significantly compared to the GRI Mech 3.0, while maintaining a a very good degree of accuracy.
• ZMN also likes to acknowledge the educational grant through the A.G. Leventis Foundation.

Figures

Fig. 18: Flame structure for CO/H2/O2/He. He/O2 = 7.0, φ = 2.0, fH2 = 1.0, Tu = 298 K, p = 10 atm. (conditions as in Fig.8 top).

Fig. 4: Laminar flame speeds of CO/H2-air mixtures using the reduced (dashed lines) and skeletal (full lines) mechanisms. Symbols: experimental results of Vagelopoulos and Egolfopoulos [48]. Tu = 298 K, p = 1 atm, XN2/XO2 = 3.76.

Fig. 3: Laminar flame speeds of syngas mixtures ( CO/H2/CH4/CO2/N2-air ) using the reduced (dashed lines) and the skeletal (full lines) mechanisms. Symbols: experimental results of Yong et al. [27]. fCH4 = 0.24 with 11% CO2 and 42.7% N2 in the fuel mixture. Tu = 298 K, p = 1 atm, XN2/XO2 = 3.76. Error bars from [27] are also shown.

Table 3: The skeletal mechanism. Units are in cm, s, mol, cal, K.

Fig. 14: Laminar flame speeds of CO/CH4/H2O/O2/N2 mixtures using the reduced (dashed lines) and skeletal (continuous lines) mechanisms. Symbols: GRI Mech 3.0 predictions. fCH4 = 5/95, Tu = 600 K at p = 1 and 10 atm.

Fig. 8: Laminar flame speeds of CO/H2 mixtures using the reduced (dashed lines) and skeletal (continuous lines) mechanisms. Symbols: experimental results of Sun et al. [33]. At p = 1 atm the oxidizer is O2,N2 with XN2/XO2 = 3.76. At p = 5, 10, 20 atm the oxidizer is O2 and He with XHe/XO2 = 7.0. Open symbols: experimental results of Singh et al. [21].

Fig. 19: Flame structure for CO/H2/CH4/H2O/CO2/O2/N2 with 25% H2O. φ = 1.0, fH2 = 5/95, fCH4 = 5/95, fCO2 = 0.5, Tu = 600 K, p = 1 atm. (conditions as in Fig.11).

Fig. 22: Ignition delay times of CO/H2/CO2/O2/N2 mixtures using the reduced (dashed lines) and skeletal (continuous lines) mechanisms. Symbols: experimental data of [28], 8.91%H2 + 11.58%CO+ 24.44%CO2 + 10.25%O2 + 44.83%N2.62

Fig. 21: Ignition delay times of CO/H2/O2/N2 mixtures ( XN2/XO2 = 3.76) for φ = 0.5 using the reduced (dashed lines) and skeletal (continuous lines) mechanisms. Symbols: experimental results of Kalitan et al. [55]. Also shown for comparison are the results with the skeletal mechanism of Boivin et al. [14] (dashed lines with ×) for the fH2 = 20/80 case.

Fig. 15: Flame structure for CO/H2/O2/N2. φ = 0.8, fH2 = 5/95, Tu = 400 K, p = 1 atm. (conditions as in Fig.5 top).

Fig. 1: Laminar flame speeds of CO/H2/H2O-air mixtures using the reduced (dashed lines) and skeletal (full lines) mechanisms. Open circles: Li et al. [31] mechanism results from [22]. Also shown are the predictions using the skeletal mechanism of Boivin et al. (dashed lines with ×) [14]. Filled symbols: experimental results of Das et al. [22]. Tu = 323 K, p = 1 atm, fH2 = 5/95, XN2/XO2 = 3.76.

Fig. 2: Laminar flame speeds of CO/H2/H2O mixtures using the reduced (dashed lines) and skeletal (continuous lines) mechanisms. Filled symbols: experimental results of Singh et al. [21]. p = 1 atm, Tu = 400 K, φ = 1, oxidiser is O2,N2 with XN2/XO2 = 3.76.

Fig. 7: Laminar flame speeds of CO/H2/CO2/O2/N2 mixtures using the reduced (dashed lines) and skeletal (continuous lines) mechanisms . Symbols: experimental results of Natarajan et al. [49]. fH2 = 5/95 and 1.0, at p = 1 atm, XN2/XO2 = 3.76 with 10% and 20% CO2 dilution.

Fig. 6: Laminar flame speeds of CO/H2 mixtures using the reduced (dashed lines) and skeletal (continuous lines) mechanisms. Filled symbols: experimental results of Singh et al. [21]. p = 1 atm, oxidizer is air.

Fig. 16: Flame structure for CO/H2/O2/N2. φ = 0.8, fH2 = 5/95, Tu = 700 K, p = 1 atm. (conditions as in Fig.5 top).

Fig. 11: Laminar flame speeds of CO/H2/CH4/H2O/CO2/O2/N2 mixtures using the reduced (dashed lines) and skeletal (continuous lines) mechanisms. Symbols: GRI Mech 3.0 results. fH2 = 5/95, fCH4 = 5/95, fCO2 = 0.5, Tu = 600 K at p = 1 and 10 atm.

Table 4: The range of fuel composition and operating conditions tested.

Table 1: The first 40 most sensitive reactions from GRI-Mech 3.0. The sensitivity analysis was conducted at Tu=323 K, φ=0.9, fH2 = 5/95 with H2O%=20%.

Fig. 10: Laminar flame speeds of CO/H2/CH4/CO2/O2/N2 mixtures using the reduced (dashed lines) and skeletal (continuous lines) mechanisms. Symbols: experimental results of Park et al. [54], p = 1 atm, Tu = 298 K. 49

Fig. 12: Laminar flame speeds of CO/H2/H2O/CO2/O2/N2 mixtures using the reduced (dashed lines) and skeletal (continuous lines) mechanisms. Symbols: GRI Mech 3.0 results. fH2 = 5/95, fCO2 = 0.5, Tu = 600 K at p = 1 and 10 atm.

Fig. 5: Laminar flame speeds of CO/H2/O2/N2 mixtures using the reduced (dashed lines) and skeletal ( continuous lines ) mechanisms. Also shown are the results using the 4- step reduced mechanism of [14] ( open squares ), the skeletal mechanism of [14] ( open circles ) from the same study, and the implementation of the skeletal mechanism of [14] in this study (dashed-dotted lines). Symbols: experimental results of Natarajan et al. [49]. fH2 = 5/95 and 1.0, at p = 1 atm, XN2/XO2 = 3.76, for Tu = 400, 500, 600 and 700 K.

Frequently asked questions

Q1. What are the contributions in "A 5-step reduced mechanism for combustion of co/h2/h2o/ch4/co2 mixtures with low hydrogen/methane and high h2o content" ?

In this study a 5-step reduced chemical kinetic mechanism involving 9 species is developed for combustion of Blast Furnace Gas ( BFG ), a multi-component fuel containing CO/H2/CH4/CO2, typically with low hydrogen, methane and high water fractions, for conditions relevant for stationary gas-turbine combustion. 

Q2. What are the future works in "A 5-step reduced mechanism for combustion of co/h2/h2o/ch4/co2 mixtures with low hydrogen/methane and high h2o content" ?

The computational results are compared to experimental measurements of the flame speeds available in the literature for a wide range of pressure, 1-20 atm., temperature, 298- 700 K and thermo-chemical conditions. The authors thank the reviewers for suggesting many validation data which helped to show the robustness of the mechanisms over wide range of conditions for flame speeds and autoignition delay times.