Article pubs.acs.org/EF
Reduction of NO by Biomass Pyrolysis Products in an Experimental Drop-Tube Hai-Sam Do,†,‡ Yutthasin Bunman,†,‡ Shiqiu Gao,† and Guangwen Xu*,† †
State Key Laboratory of Multi-phase Complex Systems, Institute of Process Engineering, Chinese Academy of Sciences, Beijing 100190, People’s Republic of China ‡ University of Chinese Academy of Sciences, Beijing 100049, People’s Republic of China ABSTRACT: To understand the mechanism of NOx reduction by reburning biomass pyrolysis products, the present work is devoted to investigate the capabilities of biomass char, tar, and pyrolysis gas (Py-gas) for NO reduction through experiments in a drop-tube reactor. Pyrolysis of distilled spirit lees at 500 °C produced the char and tar tested, while Py-gas was prepared from cylinder gases according to the analysis of experimental Py-gas. The realized NO reduction efficiency varied with the stoichiometric ratio (SR), reaction temperature, residence time, and initial NO and CO concentrations. At a specified mass rate of 0.15 g/min, tar enabled the best NO reduction in comparison to char and Py-gas. The presence of CO in Py-gas inhibited the homogeneous NO reduction reactions to cause lower NO reduction efficiency by this agent. For biomass char and tar, their realized NO reduction reactions were promoted by high temperature and high initial NO and CO concentrations. Moreover, the suitable SR values for obtaining the highest NO reduction efficiency by reburning pyrolysis products were found to be 0.6−0.8.
1. INTRODUCTION Distilled spirit lees (DSL), which is a biomass waste generated from the production process of distilled spirits, amounted to 20 million tons per year in China.1 DSL can be burned to produce steam, which not only limits the pollution as a result of DSL disposal but also offers a part of energy required by distilled spirits production.2 Considering its relatively high N content (about 3−5 wt % on a dry basis), burning DSL via the traditional way has to release high NOx emission. To reduce NOx emission in combusting this kind of high-N biomass waste, we have developed a circulating fluidized-bed decoupling combustion (CFBDC) technology that has been proven to be effective in lowering 50% NOx emission in comparison to traditional combustion.2−4 The CFBDC system is based on a dual-bed system and is composed of a fluidized-bed pyrolysis reactor and a riser combustor, as conceptualized in Figure 1. The technology separates the combustion process into pyrolysis of biomass fuel and combustion of char and volatile products, including noncondensable pyrolysis gas (Py-gas) and tar. Volatile is sent to an
intermediate position of the riser combustor to allow for its coburning with char from the riser bottom. In this way, a reburning zone is actually formed in the riser of a CFBDC system to, thus, lower the NOx emission. The lowered NOx emission should be due to the combined actions of char, tar, and Py-gas upon reduction of NOx formed by burning char N in the riser bottom. Many researchers5−8 have regarded reburning as an effective way for NOx removal in combustion. In CFBDC, the reduction of NOx by reburning of pyrolysis products involves both heterogeneous reactions between char and NOx and homogeneous reactions between volatiles (tar and Py-gas) and NOx. In recent years, a lot of studies have demonstrated the significant effect of chars9−11 and reducing gas12−14 on NOx reduction. For CFBDC, however, it is unclear what is the dominant mechanism for NOx reduction among the reaction of NOx with char, tar, and Py-gas. Tar with non-condensable Py-gas must take part in the homogeneous NOx reduction, but it has gained little attention in the past. Luo et al.15−18 found that tar promoted NO reduction by biogas, while the effectiveness of NO reduction by model tar compounds was also investigated. This study follows our preliminary finding about the high reactivity of DSL-derived tar compared to DSL-derived char in reducing NO in a micro-fluidized bed reactor.19 Accordingly, DSL-derived tar was suggested as an attractive reagent for lowering NOx emission in the CFBDC process. We intend to evaluate the ability of Py-gas among three DSL pyrolysis products (char, tar, and Py-gas) in reducing NO. Thus, the characteristics of NO reduction using char, tar, and Py-gas from DSL pyrolysis as reagents were investigated in a drop-tube reactor (DTR) to further understand the low-NOx mechanism
Figure 1. Principle conception of circulating fluidized-bed decoupling combustion (CFBDC).
Received: January 5, 2017 Revised: March 7, 2017 Published: March 8, 2017
© 2017 American Chemical Society
4499
DOI: 10.1021/acs.energyfuels.7b00040 Energy Fuels 2017, 31, 4499−4506
Article
Energy & Fuels
Figure 2. Schematic diagram of the experimental system.
Table 1. Proximate, Ultimate of DSL Pyrolysis Products and Composition Analyses of Model Py-gas proximate analysis (wt %, dry basis)
ultimate analysis (wt %, dry and ash-free basis)
sample
volatile
ash
FC
C
H
N
O+S
DSL char tar Py-gas (N2 and CO2 free) model Py-gas (vol %)
63.93 16.37
15.16 37.92
20.91 45.71
CO 27.00
CO2 12.97
6.17 2.56 7.09 6.49 C2H6 0.682
3.50 3.61 5.16
H2 6.00
52.66 67.24 71.32 49.46 CH4 11.97
37.67 26.59 16.43 44.05 N2 41.29
of CFBDC and figure out the dominant NO reduction reactions for CFBDC.
C3H8 0.089
cooling to room temperature in N2, the char sample was ground and sieved to gain a particle size range of 0.05−0.1 mm suitable for experiments. The tar was collected through a condenser system in the pyrolysis apparatus, and then it was dried over anhydrous MgSO4. The gas chromatography−mass spectrometry (GC−MS) results shown in Figure 3 revealed that phenol and phenolic compounds are the dominant chemicals in tar, in good agreement with other literature studies.20−22 The non-condensable gas from a batch pyrolysis of DSL was collected to analyze its composition in a micro gas chromatograph
2. EXPERIMENTAL SECTION 2.1. Facility and Reagents. The adopted DTR system enables not only flexible residence time control but also rapid heating of the reagent, including biomass pyrolysis char, tar, and Py-gas. The experiment was carried out with continuous feeding of a NO reduction reagent, thus simulating the conditions of reburning involved in CFBDC to the possible greatest extent. Figure 2 shows that the DTR system consisted mainly of a reaction zone (2), a preheating zone (3), a simulated flue gas supplying system (12), a few reagent feeding systems required (1a, 1b, and 8), a heating control system (7), and a flue gas analyzer system (10 and 11). The reaction zone was constructed from a corundum tube of 100 mm in inner diameter and 1680 mm in length. Five sampling ports marked as N.1−N.5 were distributed vertically along the axial direction of the reaction zone to take flue gas for measuring its composition. The outlet and inlet concentrations of NO, CO, and O2 at all such sampling points were continually monitored using a SDL M3080 online gas analyzer (Beijing SDL Technology, China), and all data were recorded in a computer. The DSL used in this study was provided by Luzhou Laojiao Group Company of China, which was pyrolyzed in a horizontal fixed-bed reactor to obtain the tar and char reagents taken for the study. The used Py-gas was a model gas made according to the composition of pyrolysis-generated non-condensable gas. Table 1 shows the results of proximate and ultimate analyses for raw DSL material and pyrolysisderived char, tar, and Py-gas. The pyrolysis of DSL was carried out at 500 °C in a N2 atmosphere for 30 min, and these ensured the complete release of volatile matter from the obtained char. After
Figure 3. Result of GC−MS analysis for the adopted tar reagent made from DSL pyrolysis. 4500
DOI: 10.1021/acs.energyfuels.7b00040 Energy Fuels 2017, 31, 4499−4506
Article
Energy & Fuels
where [NO]out is the NO concentration measured at the sampling port and [NO]in is the NO concentration at the inlet of the reaction zone, which was calculated from
(Agilent 3000). It was shown that the Py-gas was composed mainly of H2, CO, CH4, C2H6, C3H8, and CO2. Thus, these gas species were used to prepare the model Py-gas reagent used for NO reduction evaluation (see Table 1). 2.2. Procedure and Analysis. The simulated flue gas consisted of 800 ppmv NO with varied concentrations of O2 and balanced N2, and the total flow rate of flue gas for experiments was 45 L/min [standard temperature and pressure (STP)]. After heating in the preheating zone of the DTR, the simulated gas entered the reaction zone. The O2 concentration in the simulated flue gas was adjusted for each test to achieve a similar reburning stoichiometric ratio (SR) for all reagents tested (Py-gas, char, and tar). The SR referred to the ratio of the adopted O2 flow rate in each experiment to the O2 flow rate required for stoichiometric combustion of the NO reduction reagent fed into the reactor. Once the gas flow and temperatures in the reactor reached their steady states, a reagent of char, tar, or Py-gas was fed continuously into the reactor by 4 L/min (STP) of N2 carrier gas to start the reduction of NO in the simulated flue gas. The outlet gas from the DTR was sampled and quenched in a water-cooled probe, and the sampled gas further passed through a dust filter before entering the flue gas analyzer. Because formed N2O and NO2 were very few (below 10 ppm), the NOx reduction efficiency was evaluated only in terms of NO reduction. As reported in early studies,23,24 ash in char can catalyze either NO reduction reactions (reactions 3−6) or some NO formation reactions. Our analysis, however, was based on the overall reduction ability of char, which includes the catalytic effect of ash. In the experiment, we, thus, did not remove ash from char, whereas ash is actually not consumed during reaction (as a catalyst). Similarly, some gas species in Py-gas such as N2 and CO2 do not have any contribution to NO reduction. Consequently, both ash and such inert gases were not considered in normalizing the feeding rate of char and Py-gas as a NO reaction reactant (not as a catalyst). In this way, the fed amount of each reagent was carefully determined, so that all three reagents had a similar normalized mass rate of reductant, say, 0.15 g/min, as summarized in Table 2. This is also the most practical way for control in real processes. The comparisons in this paper are all based on such a mass flow rate of reagent.
⎞ ⎛ Q 0 ⎟⎟ × [NO]0 [NO]in = ⎜⎜ ⎝ Q 0 + Q1 ⎠
3. RESULTS AND DISCUSSION 3.1. NO Reduction Varying with the SR. Literature studies10,15,25 have reported that the reburning SR plays a key role on the realized NO reduction. Figure 4 shows the variation
Figure 4. Variation of NO reduction efficiency with SR at the typical reaction temperature.
with SR of realized NO reduction by char, tar, and Py-gas reagents at a reaction temperature of 900 °C in the experimental DTR. This temperature is typical for the char combustor of a CFBDC system.2 As we know, the SR values below and above 1.0 can be classified as fuel-rich and fuel-lean conditions, respectively. In Figure 4, plotted ηE enabled by tar and Py-gas reagents exhibited a variation trend of first increasing and then decreasing with raising SR. On the other hand, the realized NO reduction by char gradually decreased with the rise of SR. The maximal ηE was achieved at SR of 0.66 by tar, indicating that NO reduction species and radicals were easily formed from decomposition of tar under fuel-rich conditions. Similarly, the optimal SR was found to be 0.89 for NO reduction by Py-gas, but there was the lowest NO reduction among three tested reagents (under the same mass flow rate of reagent, say, 0.15 g/min). Thus, for a CFBDC system treating DSL, it should be operated under the fuel-rich conditions (SR < 1.0 or insufficient O2 for full combustion) to achieve the highest NO reduction through reburning of DSL pyrolysis products. Panels a, b, and c of Figure 5 show the NO reduction efficiencies varying with SR at reaction temperatures of 800, 900, and 1000 °C for reagents char, tar, and Py-gas, respectively. The achieved NO reduction generally increased with an increasing temperature for all reagents, but different temperatures caused more or less different NO reduction variations with SR. The effect of SR on realized NO reduction was obviously slight at 800 °C but became apparent at rather high temperatures. Raising the temperature increased not only the activity for NO reduction of each reductant but also
Table 2. Feeding Rates of Char, Tar, and Py-gas for NO Reduction Efficiency Comparison sample inert compound calculated feeding rate normalized mass rate CH/NO molar ratio C/NO molar ratio
char
tar
Py-gas
ash (37.92 wt %) 0.24 g/min
0.15 g/min
N2 and CO2 (54.26 vol %) 0.34 L/min (STP)
0.15 g/min
0.15 g/min
0.15 g/min
7.6
12.1
9.9
5.2
5.5
3.8
On the other hand, in fuel reburning, the actual NO reductants are C*, CHi, H, and many others generated from the C and H elements in the fed reagent. For reference, Table 2, thus, lists also the corresponding molar ratios of total C and H elements in reagent to fed NO (CH/NO ratio) and total C to NO (C/NO ratio). Both of the ratios were calculated on the basis of the ultimate analysis data shown in Table 1. While the CH/NO ratios are about 10.0, the C/NO ratios are 3.8−5.0, to ensure sufficient reduction of NO.19 The tar reagent had the highest CH/NO and C/NO ratios, but char and Py-gas had the lowest CH/NO and C/NO ratios, respectively. The NO reduction efficiency (ηE) was estimated as ⎛ [NO]out ⎞ ηE = ⎜1 − ⎟ × 100% [NO]in ⎠ ⎝
(2)
where [NO]0 is the NO concentration measured at the inlet of the preheating zone and Q0 and Q1 are the flow rates of flue gas and carrier gas (for carrying reagent), respectively.
(1) 4501
DOI: 10.1021/acs.energyfuels.7b00040 Energy Fuels 2017, 31, 4499−4506
Article
Energy & Fuels 2C* + O2 → 2C(O)
(3)
2C(O) → CO2 + nC*
(4)
2C* + 2NO → N2 + 2C(O)
(5)
2C(O) + 2NO → N2 + 2CO2 + Nc*
(6)
With the increase of SR, the raised oxygen concentration must cause a fraction of C to burn out and to form CO2 rather than to create free-active sites C* and surface complexes C(O). Also, the C−O reaction is several orders faster than the C−NO reaction,9,30 and the acceleration of char burning with raising the temperature is greater. This thus hinders the C−NO reaction, so that the more significant decrease in NO reduction at high temperatures is due to the more increased char burning. For tar and Py-gas reagents, known as the released volatiles of DSL pyrolysis, they reduce NO via the following homogeneous gas-phase reactions:8,13,25,31 tar N → HCN → NH i
(7)
NH i + NO → N2 + ...
(8)
CH i + NO → HCN + ...
(9)
HCCO + NO → HCN + ...
(10)
HCN + NO → N2 + ...
(11)
Figure 5c shows that elevating SR caused the achieved NO reduction efficiency by Py-gas to gradually increase first and decrease later, so that there was a maximum of NO reduction efficiency. The maximal ηE by Py-gas obtained at reaction temperatures of 800, 900, and 1000 °C was 19, 29, and 42% at the corresponding SR values of 1.0, 0.89, and 0.77, respectively. The gas-phase NO reduction reactions require oxygen radicals to initiate such chain reactions that generate a lot of reduction radicals, such as CHi, HCCO, etc.32 These radicals, in turn, react with NO present in flue gas via the reactions 9−11 to implement NO reduction. With low SR values, the generation of reduction radicals was enhanced by increasing the oxygen amount to have, thus, the gradually elevated NO reduction. When SR was high enough, oxidation of the reagent has to excessively occur to convert Py-gas into CO or CO2. Few reduction radicals were then generated to decrease, thus, the achieved NO reduction.33 Dagaut et al. have reported similar findings using acetylene,34 propane,35 and biomass Py-gas13 as reactants to reduce NO. In these literature studies, the maximum of achieved NO reduction was also observed at SR only slightly below 1.0 (stoichiometric). Figure 5b shows the similar variation features with SR, as in Figure 5c, for the NO reduction efficiency realized by tar reagent. Nonetheless, the realized NO reduction by tar was higher than that by Py-gas in the whole range of tested SR. At 800, 900, and 1000 °C, the maximal ηE was 24, 58, and 87% at SR values of 0.9, 0.66, and 0.55, respectively. These SR values are smaller than those required for obtaining the maximal ηE for reagent Py-gas, indicating less oxygen needed for achieving maximal ηE by tar. When reaction temperature is raised, the NO reduction by tar tended to reach its maximum at lower SR. This trend is observed also for the Py-gas reagent in Figure 5c, but the phenomenon is rather clearer for the tar reagent in Figure 5b. The general result is that the higher the temperature, the more fuel-rich condition is needed for maximizing homogeneous NO reduction.
Figure 5. Variation of NO reduction with SR at different temperatures for reagents: (a) char, (b) tar, and (c) Py-gas.
accelerated the oxidation of the reagent.25,26 The competition between NO and oxygen for their reactions with the reagent caused such results, and a deep analysis is presented below for a clear understanding. From Figure 5a, one can see that, with an increasing SR, the NO reduction efficiency realized by char decreased and the decrease became more significant at higher temperatures. It is well-known that the heterogeneous NO reduction by char requires the fuel-rich condition,10,11 so that NO is reduced to N2 by reacting with free-active sites C* and surface complexes C(O) on the char surface through the following reactions 3−6:27−29 4502
DOI: 10.1021/acs.energyfuels.7b00040 Energy Fuels 2017, 31, 4499−4506
Article
Energy & Fuels
7.6, 12.1, and 9.9 for char, tar, and Py-gas, respectively. Figure 7 correlates the CH/NO ratios and achieved NO reduction
The GC−MS data in Figure 3 clarify that phenols and phenolic compounds were the main components of DSLderived tar. Under reburning conditions, tar species, such as phenol, crack first to form simple hydrocarbon radicals, which, in turn, react with nitrogen species via reactions 7−11 to reduce NO.36 This process requires more energy and is quite complex compared to Py-gas.37 At 800 °C, the thermal cracking of tar is not extensive enough to generate many reduction radicals, so that more oxygen is needed to enhance the oxidative cracking reactions. When the temperature is up to 900−1000 °C, tar thermal cracking appears sufficient to generate enough radicals, even at low oxygen concentrations. Consequently, the maximal NO reduction efficiency appeared at a lower SR value at a higher temperature. However, the excessive oxygen (high SR) must consume reducing radicals through oxidation reactions. This competes with the NO reduction reactions requiring radicals to hinder NO reduction. A low oxygen flow rate then has to be required to increase the radical pool, rendering the maximal NO reduction at lower SR values for higher temperatures.15 3.2. NO Reduction Varying with the Reaction Temperature and Time. Figure 6 shows the achieved NO reduction
Figure 7. Correlating NO reduction efficiency realized by char, tar, and Py-gas with their CH/NO ratio (molar ratio of C and H to NO).
efficiencies at temperatures of 800, 900, and 1000 °C (from Figure 6) for all tested reagents. The hydrocarbon radicals from tar were obviously more (higher CH/NO ratio) than those from char or Py-gas, further explaining the higher NO reduction achieved by tar.19 The effect of the residence time was studied at 900 °C and a reburning SR of 0.66, and the tested range of residence time was 0.6−2.9 s. Figure 8 shows that increasing the residence
Figure 6. NO reduction efficiency varying with the reaction temperature for all three reagents.
efficiencies at different temperatures of up to 1050 °C under a typical fuel-rich condition (SR = 0.66) for three reagents. When the temperature increased from 750 to 1050 °C, ηE by char, tar, and Py-gas totally increased 44.9, 67.6, and 33.7%, respectively. Thus, a high temperature is beneficial to NO reduction for all tested reagents. At 750−850 °C, however, there was only a 9.8% increase in ηE for tar, about 1.7 times lower than that by char. The tar reagent had the best NO reduction, with its ηE increased by 57.8% at 850−1050 °C. ηE by char and Py-gas, on the other hand, only increased by 27.9 and 29.8%, respectively. These suggest that the tar reagent for NO reduction was more sensitive to the temperature than char and Py-gas. The fact would be true that, with raising the temperature, tar is easier compared to char and Py-gas to decompose and form more reductive species or radicals, such as active benzene rings, OH−, H+, and many others.15−17 Thus, in Figure 6, the variation of NO reduction efficiency with the temperature is the most distinctive for tar in comparison to that for char and Py-gas. Concerning the molar ratio of total C and H elements in reagent to fed NO, as shown in Table 2, the tar reagent has the highest CH/NO ratio. The corresponding CH/NO ratios were
Figure 8. NO reduction efficiency varying with the residence time in reaction zone at the typical reaction temperature.
time facilitates NO reduction but the NO reduction profiles varied much with the type of reagent. For the char reagent, the NO reduction efficiency quickly increased in the first 1.2 s and then increased slowly. A slow increase tendency in NO reduction was observed for both tar and Py-gas reagents in the entire range of tested residence time. These indicate that the gas-phase homogeneous reduction of NO by gaseous tar or Pygas was faster than the gas−solid heterogeneous reduction of NO by char. Thus, in short reaction time, the homogeneous NO reduction should be most important, whereas heterogeneous NO reduction by char would become dominant in a long reaction time, for example, around 1.2 s. This mechanism has also been shown by Cancès et al.7 and Shu et al.10 by reporting 4503
DOI: 10.1021/acs.energyfuels.7b00040 Energy Fuels 2017, 31, 4499−4506
Article
Energy & Fuels
condensable Py-gas) must be in atmospheres containing CO at hundreds of parts per million by volume (ppmv) to a few percents. Figure 10 shows the measured outlet CO
a quick homogeneous reduction at a residence time below 0.6 s but late start of heterogeneous reduction. Figure 8 clarifies also that the NO reduction efficiency by all reagents varied little (