Article pubs.acs.org/EF
Influence of Elevated Pressure on the Torrefaction of Wood David Agar,*,†,‡ Nikolai DeMartini,† and Mikko Hupa† †
Johan Gadolin Process Chemistry Centre, Laboratory of Inorganic Chemistry, Åbo Akademi University, FI-20500 Turku, Finland Department of Chemistry, University of Jyväskylä, P.O. Box 35, FI-40014 Jyväskylä, Finland
‡
ABSTRACT: The use of pressurized reactors in industrial processes can improve efficiency and economics. Torrefaction is a partial pyrolysis of lignocellulosic biomass designed to result in a solid product with improved fuel properties for utilization in combustion and gasification. In this work, the influence of elevated pressure on the torrefaction of wood has been investigated. Wood samples were torrefied using a pressurized thermogravimetric reactor (PTGR) with pressures of 0.1 to 2.1 MPa. The results indicate that reactor pressure, particle size of feedstock, and wood species are all factors in torrefaction yield improvements. Torrefaction at 2.1 MPa pressure improved the higher heating value (calculated) of single-particle beech cylinders from 20.4 to 22.2 MJ kg−1, the increase ranging from 7.5 to 19% from the untreated heating value. Decomposition reactions were accelerated with pressure so that a given mass yield was realized in a shorter time. At 2.1 MPa pressure and 280 °C the time was reduced by over 60% for milled aspen compared to the run made at 0.1 MPa. A key finding is that the sample torrefied at a higher pressure, but shorter residence time, despite the same mass yield had a greater carbon yield and thus also a higher energy yield. For milled aspen, most of the observed pressure-dependent effect occurred within an initial pressure increment from atmosphere up to 0.5 MPa. The findings presented will have implications for the industrial production of torrefied fuels.
1. INTRODUCTION Wood fuels have always been of great importance, and in recent years the torrefaction of wood has attracted interest as a method of producing a new generation of solid fuel. Wood is a natural material with a complex structure and composition. The main components of wood can be categorized as cellulose, lignin, hemicellulose, extractives, and ash.1 Torrefaction is a partial pyrolysis of lignocellulosic biomass in an inert atmosphere. Torrefaction is achieved using a temperature range of 200−300 °C with a relatively low rate of heating.2 At lower pyrolysis temperature, the majority of the mass loss comes from degradation of hemicellulose, while cellulose degradation becomes more significant above 260 °C. 3 Torrefaction is an emerging industrial process, whereby biomass is upgraded to a more suitable fuel for use in pulverized fuel combustion or gasification. Academic studies have been carried out on many different types of biomass,4,5 but industrial work is focused on the use of wood as a feedstock. Several observed changes take place in wood through torrefaction. Three of these are especially relevant for a fuel to be used in cofiring. First, wood becomes much more friable.6−9 Second, torrefaction raises the relative carbon content thereby increasing heating value.10,11 Third, a decrease in the concentration of carboxylic sites, due to a decomposition of hemicellulose, contributes to a reduction in the equilibrium moisture content (EMC) of torrefied wood.11,12 In the 1980s, Bourgeois and Doat had an interest in torrefaction as an industrial process for producing a more energy-efficient fuel to replace wood charcoal as well as a suitable reducing agent in metallurgy and gasification fuel.13 Today however interest in torrefied wood stems from its potential as a pulverized fuel − to offset or replace fossil coal use − and mitigate carbon dioxide emissions. Currently several technology developers are working to commercialize torrefied fuel production. Many industrial thermochemical processes utilize elevated pressure to enhance economic attractiveness. Pressurized © XXXX American Chemical Society
reactors can increase the rate of chemical reactions leading to improved efficiency and smaller size while enabling higher throughput of products. Existing pilot and semi-industrial-scale torrefaction reactors, most of which are designed as a continuous process, operate at or slightly above atmospheric pressure. A study of high-pressure torrefaction is of interest and may show potential for industrial-scale torrefied fuel production. Past investigations of wood pyrolysis at elevated pressure are almost exclusively above 300 °C. The influence of pressure on torrefaction might be expected to be similar to that on pyrolysis at higher temperature; for example, charcoal yields can be improved using elevated pressure.14 In a 1983 two-paper series, Mok et al. report on the effect of pressure and volatile residence time on char formation of wood, cellulose (microcrystalline), and lignin during pyrolysis in steam at 700 °C. Char yields from cellulose increased from 9 to 15% using pressures of 0.1 to 0.5 MPa. Importantly, experiments using fixed 0.5 MPa pressure showed that increasing the residence time of the volatiles from 1 to 18 s led to further increases in char formation from 15 to 23%.15 Using a differential scanning calorimeter with pressure vessel the authors determined the influence of pressure and vapor-phase residence time on the heat of pyrolysis. They found that high pressure and low flow increased char formation and decreased heats of pyrolysis for cellulose because condensation reactions can be exothermic.16 Blackadder et al. studied the effect of pressure on pyrolysis of wood, cellulose, and lignin in nitrogen with a pressurized thermobalance.17 They found the greatest pressure dependence for cellulose whose char yield increased from 6 to 15% through pyrolysis (750 °C) from atmosphere up to 4 MPa pressure. A less pronounced effect was observed using birch wood for which the Received: June 17, 2015 Revised: February 2, 2016
A
DOI: 10.1021/acs.energyfuels.5b01352 Energy Fuels XXXX, XXX, XXX−XXX
Article
Energy & Fuels
and partial pressures, leading to longer gas residence times and more time for secondary reactions.23 Therefore, as pressure increases, residence times also increase. It follows then that the dominating factors in char-forming reactions are pressure within the reactor, particle size, and contact between the tars and char, which is a function of particle size, reactor design, and flow rate.23 This study investigates the influence of elevated pressure on the torrefaction of wood. The influence of pressure on solid mass, energy, and carbon yields of torrefaction and on the higher heating value (calculated) is investigated. The ultimate aim of this work is to answer the following question: can torrefaction using elevated pressure be beneficial for industrial production of torrefied fuels.
char yield increased from 21 to 28% over the same pressure range. For lignin (acid type) however the pressure dependence was weak. The authors found that most of the pressuredependent effect took take place during the initial pressure step from atmosphere up to 0.5 MPa. This finding was consistent with results by Mok et al., who observed a comparable yield increase despite using a significantly lower pressure range (0.1 to 0.5 MPa).15 Although hemicellulose pyrolysis was not studied by Blackadder et al., their TG curves for wood and cellulose show a clear influence of pressure on solid yield above approximately 320 °C.17 Below this temperature, the region of hemicellulose decomposition, the effect of pressure in their data is not discernible. These findings and others, as they apply to wood charcoal production, were reviewed methodically by Wang et al.18 They concluded that elevated pressure enhances solid yield, but, due to the variable nature of wood, only fixed-carbon yield has any meaning when comparing solid pyrolysis products. They also concluded that particle size and flow conditions are key factors in fixed-carbon yield. Wang et al. confirm observations by Mok et al. − that the influence of vapor-phase residence time on charcoal formation is at least as important as that of primary pyrolysis reactions at elevated pressure. Char yields in wood pyrolysis also depend on the initial mass of the sample and on sample holder geometry which affects flow resistance encountered by vaporphase products.19 Based on the aforementioned studies, the pyrolysis behavior of wood at high temperature and pressure is well established. Below 300 °C, however, there is a lack of results at torrefaction temperatures. Only three studies using elevated pressure have been described. One of these deals exclusively with wet torrefaction and is not relevant to the present work.20 Wannapeera et al.21 recently investigated torrefaction of a tropical wood in a batch reactor using elevated pressures of 1 to 4 MPa, temperatures of 200−250 °C, and a torrefaction time of 30 min. They found that the total carbon content increased progressively with torrefaction temperature and reactor pressure from 50.1 to 62.3%. The greatest calculated higher heating value (HHV) was 25.8 MJ kg−1 from torrefaction at 250 °C and 4 MPa. This translates to a HHV increase of 26.5% corresponding to a mass and energy yield combination of 74.4 and 94.1%, respectively. No general trend in the effect of pressure on solid mass yield was reported however. The authors used a batch-type reactor with milled samples (150 mg of particle size less than 75 μm). Rapid water-bath cooling of the reactor affected mass yields due to the condensable volatiles present in torrefaction gases. Nhuchhen et al. studied torrefaction of cylindrical poplar samples (50.5 mm, ⌀ 19 mm) in N2 in a batch reactor at pressures up to 0.6 MPa.22 They report a HHV and mass yield range of 19.5−26.0 MJ kg−1 and 56−92%, respectively, using torrefaction temperatures of 220−300 °C and 15−35 min residence times. The deposition of tars on the torrefied wood was also observed in this study, and the amount of deposited tars increased with pressure. In the above studies, no flow through the reactors meant that vapor-phase pyrolysis products remained in contact with the solid sample. As a consequence, the influence of pressure and vapor-phase residence time on solid yields cannot be differentiated in their results. Since torrefaction reactors will most likely be flow through, there is a need for data from a pressurized system with flow. Even in continuous-flow reactors, as pressure increases, vapor-phase products evolving from a solid are compressed to smaller volumes. This raises their concentration
2. MATERIALS AND METHODS 2.1. Sample Materials. The wood samples used in this study are from three temperate wood species: European aspen (Populus tremula), Scots pine (Pinus sylvestris), and beech (Fagus sylvatica). These samples are referred to as “aspen”, “pine”, and “beech”, respectively. These three samples were milled and sieved to have a particle size range of 125−250 μm and mass of 50 ± 0.1 mg. In addition to the three milled sample types, single-particle beech cylinders (cut from commercially available wood doweling) were also used. The beech singleparticle cylinders had a diameter of 5 mm, an approximate length of 10 mm, and a mass of 90 ± 1 mg. The curved surface of the cylinder is smooth, while the ends are slightly rougher, being sawn and sanded. 2.2. Torrefaction in Pressurized Thermogravimetric Reactor. Torrefaction was carried out using a custom-built pressurized thermogravimetric reactor (PTGR, manufactured by Deutsch Montan Technologie fur Rohstoff, Energie, Umwelt e.V.) shown in Figure 1.24 The device is designed for determination of weight changes at pressures up to 10 MPa and temperatures up to 1100 °C.
Figure 1. Pressurized thermogravimetric reactor. Reprinted with permission from ref 24. Copyright 1998 Elsevier Science Ltd.
The two torrefaction temperatures used in this study were 240 and 280 °C and four different pressures: 0.1, 0.5, 1.0, and 2.1 MPa. Nitrogen was used as inert gas with flow rates being 2.0 L min−1 at 0.1 and 0.5 MPa pressure and 3.0 L min−1 at 1.0 and 2.1 MPa. These two flow rates were used in order to maintain pressure stability in the reactor. PTGR runs at 240 °C were carried out at only two reactor pressures: 0.1 and 2.1 MPa. All B
DOI: 10.1021/acs.energyfuels.5b01352 Energy Fuels XXXX, XXX, XXX−XXX
Article
Energy & Fuels
CHN sample preparation approximately 1.5 mg of vanadium pentoxide (V2O5) was added to each sample to ensure complete oxidation. Using a razor blade, the outer layers of the torrefied wood cylinders samples were separated from the inner material (Figure 2). The outer layer that was cut off was approximately 1.5 mm
runs were 30 min except one run with milled aspen carried out at 280 °C and 2.1 MPa for 14 min instead of 30 min. These times are counted from the moment the sample is lowered into position in the reactor. The milled sample material was loaded into a sample holder and placed within the sample lock chamber located above the reactor at room temperature. The milled wood is held in the sample holder between a tube and a net so that there is a thin layer of biomass. The holder is attached to the filament of a microbalance-winch arrangement. The wood-cylinder samples were suspended on a platinum wire of 0.4 mm diameter and lowered into the reactor using the same winch system. The cooled sample lock chamber is continuously flushed with helium gas, keeping the sample cool while the reactor is brought up to temperature. Once the reactor temperature and pressure were stable, the sample was lowered directly into the heated reactor. The reactor temperature was measured by a thermocouple located a short distance beneath the sample. Sample mass and reactor temperature was recorded every two seconds. Due to the lowering of the sample into the reactor the time required before the first weight was recorded was approximately 28 s. Three runs were carried out at each reactor pressure, and they showed good reproducibility. For example, the runs using aspen at 280 °C and 0.1 MPa pressure had dry mass yields of 75.7, 76.7, and 77.9%. The minimum and maximum dry mass yields from the three runs are within 1.1% of the average yield, which was the largest variation observed in the PTGR runs. The inert gas stream within the PTGR flows upward, producing a parallel drag force on the sample carrier. The magnitude of the drag force was taken as the difference in the sample mass recorded by the PTGR in the reactor at the end of the run and the sample mass measured on a laboratory scale at the end of the run. To correct data values this mass was added to that measured during the PTGR runs. 2.3. Treatment of Thermogravimetric Data. PTGR data consisted of sample mass, temperature, and time. The initial sample mass is taken to be the first obtained weight measurement. Mass loss during the lowering of the sample into the reactor was approximately 4 and 5.5%, respectively, for milled and single-particle samples. Mass yields were calculated using the initial dry mass of the samples before being placed in the reactor and the final mass measured on a laboratory scale after the run. The presented mass-loss curves are averages from three separate PTGR runs at the same conditions. Aspen torrefied for 14 min at 280 °C was the exception to this, and data was obtained from only a single run. The error bars in the mass yield results represent the highest and lowest values of the three replicate runs. Derivative curves were obtained using the following steps: 1. Mass-loss data was divided by the initial mass of the sample to obtain a mass fraction 2. The average was taken of mass-fraction data from the three PTGR runs. 3. Every fifth data point was selected to smooth the data. 4. The first derivative was taken of the data. 2.4. CHN Analysis. CHN analysis of wood samples was done using a FLASH 2000 Series Organic Elemental Analyzer manufactured by Thermo Scientific. Sample sizes ranged from approximately 1.4 to 2.3 mg for all measurements. Averages of the measurements performed in triplicate are presented in the paper. The oxygen content was calculated based on the difference between the sum of other elements and 100%. The ash content of the samples is extremely low and was not determined. During
Figure 2. Front and top view of wood cylinder showing outer and inner wood material and cut lines (top view). Relative dimensions of cylinder are approximate and not to scale.
thick at the center. The inner layer was approximately 2 mm × 2 mm × 7 mm. The thickness of this outer layer was a practical issue of working with a small sample and a razor blade. CHN analysis was carried out on both the outer and inner layers in order to investigate if there were any differences in the carbon content between the two layers. 2.5. Determination of Sample Moisture Content. To accurately determine CHN content, the moisture content of the samples was measured at the time of sample preparation, and water (H and O content) was subtracted out of the results to report CHN values on a dry basis. The oxygen content of the samples was calculated by the difference. The moisture content was determined by drying the sample overnight in an oven at 105 °C. The water content ranged from 1.7 to 2.3% (wet basis) for torrefied samples, and the as-received moisture contents of untreated aspen, pine, and beech were 5.5, 5.0, and 5.4% (wet basis), respectively. 2.6. Calculation of Higher Heating Value. The calculation of higher heating value was done using an empirical relation (eq 1) derived by Friedl et al. from a partial least-squares (PLS) regression model.25 They estimate a standard error of prediction of 0.36 MJ kg−1 for the average HHV values used. This relation was derived for different types of plant material all of which were untreated. HHVPLS (kJ kg −1) = 5.22C2 − 319C − 1647H + 38.6CH + 133N + 21028
(1)
The relative increase in higher heating value ΔHHV is calculated using eq 2 in which the subscript “T” refers to the torrefied material and “O” to the untreated. ΔHHV (%) = 100(HHVT − HHVO)(HHVO)−1
(2)
The mass yield of torrefaction YM was calculated using eq 3. YM (%) = 100 (dry mass of torrefied sample)(dry mass of untreated sample)−1
(3) C
DOI: 10.1021/acs.energyfuels.5b01352 Energy Fuels XXXX, XXX, XXX−XXX
Article
Energy & Fuels
Figure 3. a. Mass fraction (on an original sample basis, 6.1% moisture) of milled aspen versus torrefaction time at 280 °C for 30 min at four reactor pressures (0.1, 0.5, 1.0, and 2.1 MPa). The mass yield at point A (0.1 MPa) is the same as that at point B (2.1 MPa) but achieved with more than a 60% reduction in torrefaction time. b. Mass-loss rates (with units of mass fraction per second) versus torrefaction time of aspen at 280 °C using four reactor pressures.
3. RESULTS AND DISCUSSION
The energy yield of torrefaction YE can be found from the mass yield and the HHV of both untreated and torrefied material. YE (%) = YM(HHVT)(HHVO)−1
3.1. The Influence of Pressure on Torrefaction of Milled Aspen at 280 °C. The mass-loss curves of milled aspen torrefied at 280 °C at four different pressures are shown in Figure 3a. The mass loss in the first 30−60 s is mostly drying as the water content was 6.1%. The first derivatives of the mass-loss curves are shown in Figure 3b with units of mass fraction per second. For all runs it can be seen that both the initial and peak rate of mass loss of aspen depend strongly on pressure. After this initial
(4)
Carbon yield YC was calculated using the results of CHN analysis. YC (%) = 100(C in torrefied sample)(C in untreated sample)−1
(5)
Eqs 1−5 are on a dry-mass basis. D
DOI: 10.1021/acs.energyfuels.5b01352 Energy Fuels XXXX, XXX, XXX−XXX
Article
Energy & Fuels
Figure 4. a. Average mass yield of milled aspen torrefied for 30 min at 280 °C at pressures from 0.1 to 2.1 MPa. The error bars represent the minimum and maximum values for the three runs. b. Average mass yield of milled aspen torrefied for 30 min at 240 °C at pressures 0.1 and 2.1 MPa. The error bars represent the minimum and maximum values for the three runs.
yields of aspen torrefied for 30 min at 280 °C decreased from 77 to 69% when increasing pressure from 0.1 to 2.1 MPa (Figure 4a). Error bars show the minimum and maximum yield from the three runs. A similar, though less pronounced, trend was observed in the torrefaction of aspen at 240 °C. The mass yields were 92 and 87% at 0.1 and 2.1 MPa pressure, respectively (Figure 4b). The mass yields of beech torrefied for 30 min at 280 °C, both in milled form and as a single-particle cylinder, are presented in Figure 5. The single-particle cylinders had a mass-yield range of 71−82%, while the corresponding range for milled beech was 68−77%.
dependence, which lasts approximately 6 min, the rates are constant and roughly equal being independent of pressure. Early decomposition reactions in torrefaction depend strongly on pressure (Figure 3b). Consequently, mass and energy yields were observed to decrease with pressure when using a constant residence time of 30 min. A given mass yield obtained at atmospheric pressure can then be obtained using higher pressure but with a reduction in torrefaction time. Referring to the 0.1 and 2.1 MPa curves in Figure 3a, the final mass yield after 30 min at 0.1 MPa (point A) is the same as the mass yield after some 11 min at 2.1 MPa (point B). Solid Yields. The mass yields varied inversely with reactor pressure. Over the pressure range of 0.1 to 2.1 MPa, the mass E
DOI: 10.1021/acs.energyfuels.5b01352 Energy Fuels XXXX, XXX, XXX−XXX
Article
Energy & Fuels
Figure 5. Mass yields of milled beech and single-particle beech cylinders as a function of torrefaction pressure at 280 °C and torrefaction time 30 min. Only one run was made for each of the conditions, so no error bars are given.
the cylinders and higher carbon retention. Both of these phenomena can be attributed to the secondary char forming reactions in the cylinders compared to the milled beech. The HHV improvement for the milled beech and the inner cylinder section was similar. The differences in the carbon yield and ΔHHV between the inner and outer cylinder sections increase with torrefaction pressure further supporting the concept that pressure resulted in increased char formation due to reactions between tars and the torrefied wood. The differences in the measured carbon content between inner and outer cylinders (Figure 6) can be a reflection of the residence time of vapor-phase products. During torrefaction, evolved vapor-phase products are swept away in the gas flow from the sample. Conceptually, the inner mass of the cylinder can be thought of as being surrounded with a wooden sheath − namely the outer cylinder (Figure 2). Vapor-phase products from within must diffuse through the wood medium which requires time. Consequently, residence time of vapors within the bulk of the sample is extended and is a function of distance from the outer surface. Longer residence time allows more so-called secondary (char-forming) reactions to take place within the solid and results in a greater carbon yield. Therefore, as observed in earlier studies,15,18 particle size is a key factor in carbon yield. Torrefaction at atmospheric pressure resulted in the same C throughout the cylinder, whereas the inner cylinder C content was measured to be 1% higher at 2.1 MPa. Observed differences between inner and outer cylinders at elevated pressure could also stem from nonuniform torrefaction throughout the bulk of the sample. Limitations in heat transfer within wood samples, as assessed using the dimensionless Biot Number, can result in a nonuniform temperature distribution.26 Studies have shown that for 5 mm diameter samples at atmospheric pressure, this effect may be negligible.27 However, for larger diameter samples, the temperature at the core of the samples can rise well above that in the reactor. A higher reactor pressure will tend to enhance this effect (increase the Biot Number) because convective and conductive heat transfer within the reactor will increase with pressure. Therefore, the increase in carbon in the inner cylinder, which becomes significant at
The mass and energy yields are shown in Table 1. The results of CHN analysis and calculations using eqs 1−5 are also presented. The carbon content as a function of the four torrefaction pressures for torrefaction at 280 °C is plotted in Figure 6. The carbon content of the aspen increased with pressure from 50.3 to 53.2%, compared to the untreated aspen which had a carbon content of 48.4%, Figure 6. The corresponding heating values (calculated using eq 1) had a range of 19.9−21.1 MJ kg−1. Torrefaction using pressures from atmosphere up to 2.1 MPa resulted in a ΔHHV of 3.9 to 10.1%. For torrefaction at 240 °C, the HHV after torrefaction was calculated to be 19.9 and 20.3 MJ kg−1 at atmosphere and 2.1 MPa pressure, respectively. The HHV of untreated aspen was 19.2 MJ kg−1. After torrefaction at 280 °C and 2.1 MPa pressure, the carbon content of the samples increased from the untreated state by 8.6% (inner cylinder), 8.2% (beech), 7.7% (outer cylinder), 5.4% (pine), and 4.8% (aspen). For aspen at 240 °C the increase was 2.8%. The increase in carbon content for beech, both milled and cylinder sections, is linear as seen by fitted linear equations. The difference in C between the inner and outer sections of the cylinders increases with torrefaction pressure. It is noted, however, that most of this difference is within the error level of C determination as given in Table 1. The relation for aspen is nonlinear. In the pressure range of 0.1−0.5 MPa, pressure results in a high C content, while, at pressures above this range, the C content increases only slightly. As noted earlier, shorter residence times can be used at higher pressures to achieve the same mass yield. One run was made with aspen at 2.1 MPa and 280 °C for 14 min to analyze the composition. The CHN values of this sample can be compared to the aspen torrefied for 30 min at 280 °C and 0.1 MPa (Table 1). High-pressure torrefaction using a shorter time produced a higher carbon (82 vs 80%) and energy (83 vs 80%) yield. Therefore, ΔHHV is improved from 3.9% at 0.1 MPa to 10.9% at 2.1 MPa, despite mass yields being roughly equal. For milled beech, YE decreased from 85 to 80% with a pressure increase from 0.1 to 2.1 MPa, whereas the beech cylinders decreased from 88 to 84% and 88 to 83% for inner and outer sections, respectively. This was due to both lower mass loss for F
DOI: 10.1021/acs.energyfuels.5b01352 Energy Fuels XXXX, XXX, XXX−XXX
Article
Energy & Fuels Table 1. CHN Analysis, Heating Values, and Yields of Torrefied Wood Samples sample
T (°C)
Pa (MPa)
untreated aspen
untreated pine
untreated beech
aspen
280
0.1
0.5
1.0
2.1
2.1c
aspen
240
0.1
2.1
pine
280
0.1
2.1
beech
280
0.1
0.5
1.0
2.1
beech cylinder, innerd
280
0.1
0.5
1.0
2.1
e
C
H
N
Ob
O/C
H/C
HHV (MJ/kg)
ΔHHV (%)
YC (%)
YM (%)
YE (%)
48.43 48.33 48.49 51.23 51.19 51.31 48.14 48.05 48.31 50.33 50.23 50.45 52.18 51.55 52.60 52.87 52.66 52.99 53.18 53.14 53.24
5.90 5.86 5.92 6.16 6.09 6.20 5.65 5.64 5.67 5.68 5.66 5.69 5.72 5.65 5.84 5.71 5.68 5.74 5.52 5.48 5.59
0.11
45.56
0.94
0.12
19.15
0.0
100.0
100.0
100.0
0.09
42.53
0.83
0.12
20.43
0.0
100.0
100.0
100.0
0.04
46.17
0.96
0.12
18.97
0.0
100.0
100.0
100.0
0.14
43.85
0.87
0.11
19.89
3.9
79.8
76.8
79.8
0.10
41.99
0.80
0.11
20.71
8.1
79.4
73.7
79.7
0.12
41.30
0.78
0.11
21.02
9.8
78.5
71.9
78.9
0.19
43.85
0.82
0.10
21.09
10.1
75.5
68.7
75.7
53.37 53.24 53.46 50.16 50.07 50.26 51.18 51.08 51.27 53.39 53.08 53.62 56.59 56.55 56.61 52.58 52.26 52.84 53.68 53.42 53.84
5.69 5.63 5.76 5.92 5.89 5.97 5.90 5.89 5.92 5.77 5.71 5.81 5.62 5.57 5.66 5.47 5.38 5.53 5.71 5.67 5.75
0.11
40.82
0.76
0.11
21.24
10.9
82.3
74.6
82.8
0.12
43.80
0.87
0.12
19.89
3.9
95.1
91.8
95.4
0.12
42.80
0.84
0.12
20.33
6.2
92.3
87.3
92.7
0.12
40.73
0.76
0.11
21.28
4.2
81.8
78.4
81.7
0.12
37.67
0.67
0.10
22.73
11.2
78.3
70.9
78.9
0.10
41.84
0.80
0.10
20.80
9.6
84.2
77.1
84.5
0.09
40.52
0.75
0.11
21.39
12.8
80.4
72.2
81.4
54.23 53.46 54.72 56.38 56.05 56.59 51.78 51.36 52.15 52.80 52.48 53.34 54.12 53.84 54.43 56.70 56.57
5.39 5.20 5.50 5.30 5.23 5.38 5.41 5.37 5.43 5.37 5.32 5.43 5.26 5.17 5.32 5.14 5.11
0.12
40.27
0.74
0.10
21.50
13.3
80.6
71.6
81.1
0.11
38.21
0.68
0.09
22.46
18.4
79.0
67.5
79.9
0.00
42.81
0.83
0.10
20.41
7.6
87.9
81.7
87.9
0.04
41.79
0.79
0.10
20.84
9.9
88.1
80.3
88.2
0.10
40.53
0.75
0.10
21.39
12.8
87.0
77.4
87.3
0.05
38.11
0.67
0.09
22.51
18.7
83.7
71.1
84.4
G
DOI: 10.1021/acs.energyfuels.5b01352 Energy Fuels XXXX, XXX, XXX−XXX
Article
Energy & Fuels Table 1. continued sample beech cylinder, outerd
T (°C) 280
Pa (MPa) 0.1
0.5
1.0
2.1
C
H
56.78 51.74 51.44 52.29 52.65 52.57 52.76 53.85 53.57 54.33 55.79 55.31 56.15
5.16 5.39 5.38 5.42 5.37 5.35 5.39 5.31 5.24 5.37 5.06 4.99 5.13
N
Ob
O/C
H/C
HHV (MJ/kg)
ΔHHV (%)
YC (%)
YM (%)
YE (%)
0.04
42.82
0.83
0.10
20.39
7.5
87.8
81.7
87.9
0.09
41.89
0.80
0.10
20.79
9.6
87.8
80.3
88.0
0.14
40.70
0.76
0.10
21.30
12.3
86.6
77.4
86.9
0.17
38.98
0.70
0.09
22.06
16.3
82.4
71.1
82.7
Reactor flow rates were 2.0 L min−1 (0.1 and 0.5 MPa) and 3.0 L min−1 (1.0 and 2.1 MPa). bCalculated by difference from 100%. cTorrefaction time was 14 min for this run. dMass yield is for whole cylinder. eThe values of CHNO are averages. The minimum and maximum values for C and H are included below the average. a
Figure 6. Relationship between carbon content and torrefaction pressure for the beech (×), inner (□) and the outer (Δ) beech cylinder, and aspen (○) samples. The samples were all torrefied for 30 min at 280 °C.
used in this study are similar in thickness to wood chips, and therefore the results with the cylinders are expected to be more representative of torrefaction of wood chips at higher pressure than that observed from milled wood results. It follows that particles larger than wood chips would exhibit even greater inhomogeneity. This may lead to the particle being cooked on the outside but overcooked on the inside. As size reduction of untreated wood requires significantly more energy than torrefied wood,4 torrefaction using a large particle size would be beneficial insofar as inhomogeneity of the particle is tolerable. For the production of pellets made from torrefied wood, a high degree of torrefaction makes pelletization more challenging and has a negative impact on the durability of pellets.28,29 Therefore, avoiding an overly torrefied core may be the key concern in this regard. Atmospheric torrefaction of milled aspen at both 240 and 280 °C produced approximately the same ΔHHV (3.9%) despite the temperature difference. However, the mass, energy, and carbon yields for the 240 °C runs are much higher (92, 95, and 95%) than the latter (77, 80, and 80%). For this wood type, this indicates that torrefaction at higher temperature is not warranted
pressures above 0.1 MPa, could also be caused by this coretemperature overshoot. In other words, the core of the sample is effectively torrefied at higher temperature. The results presented in Figure 6 indicate that the chemical changes brought about in torrefaction are neither uniform throughout the bulk of a single-particle cylinder nor between different wood species. The two-step nonlinear behavior of milled aspen compared to milled beech suggests that the observed differences may be species-dependent − since both samples had the same initial mass and the same sample holder was used. Two different reactor flow rates were used in this study. The flow rate was not found to clearly affect mass loss. One flow rate was used for the experiments at 0.1 and 0.5 MPa pressure and a second for experiments at 1.0 and 2.1 MPa pressure. The results for each sample type within these two flow ranges are due to changes in the only independent variable: pressure. The results in Figure 6 have implications for an industrial torrefaction process at higher pressures. If the particle size is large, the carbon content throughout the particle is less homogeneous than for small particles. The wood cylinders H
DOI: 10.1021/acs.energyfuels.5b01352 Energy Fuels XXXX, XXX, XXX−XXX
Article
Energy & Fuels
part of the activities at the Johan Gadolin Process Chemistry Centre, a Centre of Excellence financed by Åbo Akademi University. This work has been partly carried out within CLIFF (2014-2017) as part of the activities of Åbo Akademi University. Support from the National Technology Agency of Finland (Tekes), Andritz Oy, Valmet Technologies Oy, Amec Foster Wheeler Energia Oy, UPM-Kymmene Oyj, Clyde Bergemann GmbH, International Paper Inc., and Top Analytica Oy Ab is gratefully acknowledged. The authors also wish to thank Peter Backman for thoughtful discussion and experimental work with the pressurized thermogravimetric reactor and Niklas Vähä-Savo and Luis Bezerra for assistance with CHN analysis.
since lower temperature produces the same improvement in HHV while retaining a much greater portion of the original feedstock. Lastly, Blackadder et al. found that char formation in wood pyrolysis was enhanced by pressure above temperatures of approximately 320 °C.17 This is in fact opposite of what is observed in wood torrefaction where mass yields decrease with reactor pressure. The results, however, show than the C content of the product is increasing with pressure, but the large amount of volatile matter still present in the sample at torrefaction temperatures is probably the reason behind the difference in behavior. As temperature is increased, there must be a temperature and pressure at which the mass yields begin to increase so that the results approach those observed by Blackadder et al.17 Determination of this crossover point and how it depends on wood species is an interesting topic of future studies.
■
4. CONCLUSIONS There are clear benefits to increased pressure during the torrefaction of wood. Increased pressure and lower temperatures and/or shorter residence times can be used for increased yields and higher carbon content resulting in higher energy yields. The rate of decomposition reactions in the torrefaction of wood is a strong function of reactor pressure. As a consequence, elevated pressure reduces the torrefaction time required to achieve a given mass yield. Additionally, torrefaction at elevated pressure with shorter residence times or for larger sample sizes improves the carbon yield of wood compared to atmospheric torrefaction. Consequently, the heating value improvement is enhanced with higher pressure. Torrefaction at low temperature and high pressure results in significantly greater mass, energy, and carbon yields than torrefaction at high temperature and low pressure. After approximately 6 min, the mass-loss rate for all wood samples was independent of pressure. It was roughly equal and constant. This shows that to minimize torrefaction time, it is sufficient that the early stages of torrefaction be carried out at elevated pressure. In principle, it might be beneficial to have a two-stage process where the first minutes are under pressure while the remaining time is at atmospheric pressure. This could have the benefit of lower overall capital costs than having the full reactor built to handle elevated pressures. More experimental data is needed on torrefaction at elevated pressure in order to obtain pressure dependence of kinetic parameters. Torrefaction at elevated pressure of other wood species and other lignocellulosic biomasses, such as straw or miscanthus, would also be a relevant topic of study. The diameter of the wood cylinders used in this work is similar to that of the narrowest dimension of a wood chip so the results are expected to be relevant to wood chips. However, this should be verified by the torrefaction of chips at elevated pressure.
■
■
ABBREVIATIONS C = carbon mass percent in dry sample (%) H = hydrogen mass percent in dry sample (%) HHV = higher heating value (MJ kg−1) HHVO = higher heating value of untreated sample (MJ kg−1) HHVPLS = higher heating value based on partial least-squares (PLS) model (kJ kg−1) HHVT = higher heating value of torrefied sample (MJ kg−1) N = nitrogen mass percent in dry sample (%) O = oxygen mass percent in dry sample (%) P = pressure (MPa) PTGR = pressurized thermogravimetric reactor T = torrefaction temperature (°C) t = torrefaction time (s) YC = carbon yield of torrefaction, dry mass basis (%) YE = energy yield of torrefaction, dry mass basis (%) YM = solid mass yield of torrefaction, dry mass basis (%) ΔHHV = relative higher heating value increase (%) REFERENCES
(1) Fengel, D.; Wegener, G. Wood: chemistry, ultrastructure, reactions; De Gruyter: Berlin, 1989 p 613 s. (2) Doat, J. Revue Bois et Forêts des Tropiques 1985, 208, 57−67. (3) Khazraie Shoulaifar, T.; DeMartini, N.; Willför, S.; Pranovich, A.; Smeds, A. I.; Virtanen, T. A. P.; Maunu, S. L.; Verhoeff, F.; Kiel, J. H. A.; Hupa, M. Energy Fuels 2014, 28, 3863−3872. (4) Agar, D.; Wihersaari, M. Biomass Bioenergy 2012, 44, 107−111. (5) van der Stelt, M. J. C.; Gerhauser, H.; Kiel, J. H. A.; Ptasinski, K. J. Biomass Bioenergy 2011, 35, 3748−3762. (6) Repellin, V.; Govin, A.; Rolland, M.; Guyonnet, R. Biomass Bioenergy 2010, 34, 923−930. (7) Arias, B.; Pevida, C.; Fermoso, J.; Plaza, M. G.; Rubiera, F.; Pis, J. J. Fuel Process. Technol. 2008, 89, 169−175. (8) Phanphanich, M.; Mani, S. Bioresour. Technol. 2011, 102, 1246− 1253. (9) Bridgeman, T. G.; Jones, J. M.; Williams, A.; Waldron, D. J. Fuel 2010, 89, 3911−3918. (10) Bridgeman, T. G.; Jones, J. M.; Shield, I.; Williams, P. T. Fuel 2008, 87, 844−856. (11) Yan, W.; Acharjee, T.; Coronella, C.; Vásquez, V. Environ. Prog. Sustainable Energy 2009, 28, 435−440. (12) Khazraie Shoulaifar, T.; DeMartini, N.; Ivaska, A.; Fardim, P.; Hupa, M. Bioresour. Technol. 2012, 123, 338−343. (13) Bourgeois, J. P.; Doat, J. In Torrefied wood from temperate and tropical species − advantages and prospects. Bioenergy 84; 1984; Vol. 84, pp 15−21. (14) Antal, M. J.; MOK, W. S. L. Energy Fuels 1990, 4, 221−225. (15) Mok, W.S.-L.; Antal, M. J., Jr. Thermochim. Acta 1983, 68, 155− 164. (16) Mok, W. S.-L.; Antal, M. J. Thermochim. Acta 1983, 68, 165−186. (17) Blackadder, W.; Rensfelt, E. A pressurized thermobalance for pyrolysis and gasification studies of biomass, wood and peat. In
AUTHOR INFORMATION
Corresponding Author
*Phone: 358 40 805 3934. E-mail: daagar@jyu.fi. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS Funding from The Fortum Foundation and Johan Gadolin Scholarship for D. Agar is gratefully acknowledged. This study is I
DOI: 10.1021/acs.energyfuels.5b01352 Energy Fuels XXXX, XXX, XXX−XXX
Article
Energy & Fuels Fundamentals of Thermochemical Biomass Conversion; Overend, R. P., Ed.; 1985; pp 747−759. (18) Wang, L.; Trninic, M.; Skreiberg, Ø; Gronli, M.; Considine, R.; Antal, M. J. Energy Fuels 2011, 25, 3251−3265. (19) Rath, J.; Wolfinger, M. G.; Steiner, G.; Krammer, G.; Barontini, F.; Cozzani, V. Fuel 2003, 82, 81−91. (20) Bach, Q. V.; Tran, K. Q.; Khalil, R. A.; Skreiberg, Ø; Seisenbaeva, G. Energy Fuels 2013, 27, 6743−6753. (21) Wannapeera, J.; Worasuwannarak, N. J. Anal. Appl. Pyrolysis 2012, 96, 173−180. (22) Nhuchhen, D. R.; Basu, P. Energy Fuels 2014, 28, 3110−3121. (23) Di Blasi, C. Prog. Energy Combust. Sci. 2008, 34, 47−90. (24) Whitty, K.; Backman, R.; Hupa, M. Carbon 1998, 36, 1683−1692. (25) Friedl, A.; Padouvas, E.; Rotter, H.; Varmuza, K. Anal. Chim. Acta 2005, 544, 191−198. (26) Prins, M. J.; Ptasinski, K. J.; Janssen, F. J. J. G. J. Anal. Appl. Pyrolysis 2006, 77, 28−34. (27) Basu, P.; Rao, S.; Dhungana, A. Can. J. Chem. Eng. 2013, 91, 466− 474. (28) Järvinen, T.; Agar, D. Fuel 2014, 129, 330−339. (29) Larsson, S. H.; Rudolfsson, M.; Nordwaeger, M.; Olofsson, I.; Samuelsson, R. Appl. Energy 2013, 102, 827−832.
J
DOI: 10.1021/acs.energyfuels.5b01352 Energy Fuels XXXX, XXX, XXX−XXX
