- Email: [email protected]
The role of ash particles in the bed agglomeration during the fluidized bed combustion of rice straw
Bioresource Technology 100 (2009) 6505–6513
Contents lists available at ScienceDirect
Bioresource Technology journal homepage: www.elsevier.com/locate/biortech
The role of ash particles in the bed agglomeration during the fluidized bed combustion of rice straw Huanpeng Liu a,b,*, Yujie Feng b, Shaohua Wu a, Dunyu Liu a a b
School of Energy Science and Engineering, Harbin Institute of Technology, Harbin, China School of Municipal and Environmental Engineering, Harbin Institute of Technology, Harbin, China
a r t i c l e
i n f o
Article history: Received 21 March 2009 Received in revised form 24 June 2009 Accepted 25 June 2009 Available online 6 August 2009 Keywords: Fluidized bed Defluidization time Agglomerate fraction Coating layer Neck
a b s t r a c t In this paper, the effects of fluidization velocity, bed temperature and fuel feeding rate on the defluidization time and agglomerate fraction in the fluidized bed combustion of rice straw were studied. The fuel ash, necks in agglomerates and coating layers of bed particles were studied by means of the scanning electron microscope, coupled with energy-dispersive spectroscopy (SEM/EDS). Results showed that the stickiness of bed particles induced by coating layers is the direct reason for bed defluidization. The alkali metals such as K and Na mainly exist in the outer layer of rice straw particles. During combustion the high temperature can cause the alkali species melting and coating the surfaces of ash particles. Consequently, ash particles become sticky and tend to adhere to the surfaces of bed particles. The large-sized ash particles may act as the necks in the formation of agglomerates. The small-sized ash particles may contribute to the formation of coating layers. Ó 2009 Elsevier Ltd. All rights reserved.
1. Introduction Biomass as a kind CO2-neutral energy source has been paid a great deal of attention (Li et al., 2008). At the same time, fluidized beds are widely used for the combustion of biomass because they are suitable for the low-grade fuels. However, in the practical operations, the bed agglomeration which can result in the unscheduled shutdown of fluidized bed reactors has become a serious problem. So, extensive studies have been carried out on it. It has been verified that the bed agglomeration is attributed to the alkali species in the biomass ash. The detection and counteraction methods have been studied (Nijenhuis et al., 2007; Vamvuka et al., 2008; Arvelakis et al., 2001; Bartels et al., 2008; Bakker et al., 2002). The effects of running parameters (such as bed temperature, fluidization velocity, etc.) on the bed agglomeration were studied by Lin et al. (2003). The ash-forming elements in the different biomass were analyzed and the ash formation mechanisms during the combustion of biomass in fluidized beds were studied too (Lind et al., 2000; Werkelin et al., 2005). It has been widely accepted that the alkali species in the biomass ash can be transferred to the surface of bed particles and result in the formation of coating layer during combustion. The coating layer consists of several superim* Corresponding author. Address: School of Energy Science and Engineering, No. 457 mailbox, Harbin Institute of Technology, Harbin 150001, China. Tel.: +86 451 86412578; fax: +86 451 86412528. E-mail address: [email protected] (H. Liu). 0960-8524/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.biortech.2009.06.098
posed layers. The innermost layer mainly contains alkali silicates, whereas the outermost layer is rich in calcium or magnesium (Nuutinen et al., 2004). Although the deposition of fine ash particles, condensation of volatile alkali species and the collisions between the bed particles and burning char particles have been reported as critical transport mechanisms (Scala and Chirone, 2006, 2008), the formation process of the coating layer has not been determined up to now. Meanwhile, studies have been conducted on the bed agglomeration mechanism. Up until now, two different types of agglomeration formation have been proposed: melt-induced agglomeration and coating-induced agglomeration. In the melt-induced agglomeration, the molten ash particles are considered to act as the necks to bond the bed particles in the agglomerate formation process (Chirone et al., 2006; Lin et al., 2003). In the coating-induced agglomeration, the melting of coating layers of bed particles is considered to be responsible for the bed agglomeration (Natarajan et al., 1998; Öhman et al., 2000, 2005; Brus et al., 2005). But the formation mechanism of the agglomerate has not been determined because there are not enough experimental evidences to verify these hypotheses. In present work, rice straw is used as the study object. The experiment is carried out in a lab-scale bubbling fluidized bed combustor. The effects of fluidization velocity, bed temperature and fuel feeing rate on the defluidization time and agglomerate fraction are studied. As for the bed agglomeration mechanism, we focus on the formation processes of agglomerates and coating layers, specially the role of ash particles in these formation
6506
H. Liu et al. / Bioresource Technology 100 (2009) 6505–6513
1. pressure meter 2. temperature controllor 3. thermalcouple 4. heating wire 5. refractory protection 6. heat insulator 7. distributor 8. flow-meter 9. Compressor 10.windbox 11.discharge pipe Fig. 1. Schematic of the experimental apparatus.
(a)
(b)
6
200
V=0.1m/s
100
F=15g/h
3
t f
80
2
60 40
1
Defluidization time t (min)
4
120
20 0 740
760
780
800
820
840
860
880
900
0 920
T=800°C
t
V=0.1m/s
f
4
125 100
3
75
2
50 1
25 0
0 10
Bed temperatureT (°C)
(c)
20
25
30
6
t
T=800°C
f F=15g/h
5
150 4
125
3
100 75
2
50 1
25
0.07
0.08
0.09
0.10
0.11
0.12
Agglomerates fraction f (%)
175
Defluidization time t (min)
15
Fuel feeding rateF (g/h)
200
0 0.06
5
150
Agglomerate fraction f (%)
140
6
200 175
5
160
Agglomerates fraction f (%)
Defluidization time t (min)
180
0 0.13
Fluidization velocity V (m/s) Fig. 2. Effects of running parameters on the defluidization time and agglomerate fraction.
processes. The SEM/EDS analysis was carried out on the outer surfaces of ash particles, the outer surfaces and cross sections of
agglomerates, as well as the outer surfaces of bed particles. Some valuable findings are obtained.
H. Liu et al. / Bioresource Technology 100 (2009) 6505–6513
2. Methods The experimental apparatus is schematically shown in Fig. 1. The reactor is made of high temperature resistant steel tube with an inner diameter of 32 mm and a total height of 600 mm. A perforated steel plate distributor is located at the bottom of bed. The reactor is heated by means of heating wire
6507
with a standard power of 2.6 kW. Quartz sand sieved to a size range 200–300 lm is used as the bed material. The sand contains >98% SiO2 and the initial mass of bed material of each test is 200 g. The bed temperature is continuously monitored and controlled by the thermocouples. Air is fed to the windbox by an air compressor. The fluidization velocity is regulated by a flow-meter.
Fig. 3. Close-up view and SEM micrographs of rice straw ash.
6508
H. Liu et al. / Bioresource Technology 100 (2009) 6505–6513
Table 1 Elemental analysis of rice straw and ash used in experiments. Moisture
Ash
Volatile
Fixed carbon
Heat value
Proximate analysis of rice straw (as received) (wt.%) 4.99 13.87
65.05
16.09
13,860 kJ/kg
Carbon
Nitrogen
Sulphur
Oxygen
1.09
0.08
36.66
Hydrogen
Ultimate analysis of rice straw (dry basis) (wt.%) 41.93 5.64 Na2O
MgO
Al2O3
Ultimate analysis of rice straw ash (wt.%) 1.98 2.47 1.94
SiO2
P2O5
SO3
K2O
CaO
Fe2O3
56.23
1.93
4.95
19.4
10.12
0.98
Rice straw used in the experiment comes from a farm in Heilongjiang province in China. Table 1 shows the proximate and ultimate analyses of rice straw and the composition of rice straw ash. Rice straw is broken to the size range 1–2 mm by a biomass crushing engine before experiments. Fuel particles are store in a container and continuously fed into the combustor by a screw feeder. Bed temperature will increase as soon as the fuel is introduced. In order to keep bed temperature at a given value, a trial test is performed before every case to determine the pre-heated temperature of bed before feeding fuel. Even so, bed temperature still fluctuates slightly around the given value. So, bed temperature is defined as the averaged value throughout the test. In present study, fluidization velocity varies from 0.067 to 1.2 m/s and fuel feeding rate varies from 10 to 30 g/h. The excess air factors for all tests are in the range 1.5–8.0. So, all tests should run under the combustion condition. The combustor start-up is accomplished by electrically heating the bed of fresh inert sand. When the bed temperature reaches the desired value, the test and fuel feeding begin. During the test, stir will be performed as soon as the bed defluidization occurs. The test is considered to be over when the bed fluidization can not be recovered by the stir. When test ends, bed materials is discharged and sieved. In this study, agglomerate fraction is defined as the weight ratio of agglomerates to the bed material. It is used to indicate the extent of agglomerate formation among tests with different running parameters. Here, the sizes of agglomerates are larger than 2 mm. Because agglomerates are very fragile, the value of agglomerate fraction is sensitive to the way of handing and collection. So, every test is repeated for three times. The agglomerate fraction under one running condition is the average of the three repeated tests. The defluidization time is defined as the time between the start of fuel feeding to the defluidization of the bed. In order to study the cross sections of the agglomerates, some samples are mounted in epoxy resin, cut, polished with sandpapers, finally, coated with gold layer for the SEM/DES analyses.
Table 2 EDS spot analyses of the rice straw ash. Points
a b c d e f g h
Elements (wt.%) Si
Na
K
Ca
Mg
O
Cl
S
P
44.97 42.61 55.87 51.92 44.87 39.79 51.41 52.06
0.97 0.73 9.84 1.00 5.01 3.17 0.49 5.49
3.47 18.88 0.97 0.67 8.60 8.16 1.3 10.84
4.55 19.12 0.64 0.0 5.23 5.0 1.0 0.0
0.18 0.58 0.0 0.42 0.62 0.78 0.35 1.05
37.54 15.54 28.79 43.33 29.93 39.76 44.24 23.22
0.47 1.76 0.0 0.32 1.99 1.21 0.0 0.0
1.04 0.0 1.48 0.72 1.18 0.75 0.18 0.44
6.81 0.77 2.41 1.64 2.58 1.38 1.03 6.79
3. Results and discussion 3.1. Effects of running parameters on the defluidization time and agglomerate fraction Fig. 2a shows the effect of bed temperature (T) on defluidization time (t) and agglomerate fraction (f) when fuel feeding rate (F) is 15 g/h and fluidization velocity (V) is 0.1 m/s. There is no obvious tendency of bed defluidization when bed temperature is below 750 °C. When bed temperature is above 750 °C, bed particles appear sticky. It has been known that the stickiness of bed particles is attributed to the melting of low-melting point alkali silicates coating layer on the surface of bed particles. Defluidization time decreases with the bed temperature increasing from 750 to 900 °C. This indicates that some compounds in the coating layer begin to melt when bed temperature is 750 °C. As bed temperature increases, the fraction of melts in the coating layers of bed particles increases. Then the viscosity of coating layer increases. The viscous forces between the bed particles increase too. So, the defluidization time will decrease with the increase of bed temperature. At the same time, agglomerate fraction increases firstly and reaches a maximum value at the bed temperature of 850 °C. Interestingly, when bed temperature increases as high as 900 °C, although bed defluidization occurs soon, agglomerate fraction decreases a little instead of increases. This indicates that the stickiness of bed particle induced by the melting of coating layers is the direct reason for the bed defluidizaton, but is not the only reason for the agglomerate formation. Fig. 2b shows the effect of fuel feeding rate on defluidization time and agglomerate fraction when bed temperature is 800 °C and fluidization velocity is 0.1 m/s. Defluidization time decreases with the increase of fuel feeding rate. It has been accepted that the alkali species which cause the formation of coating layers come form the biomass ash. So, with fuel feeding rate increasing, the amount of alkali species in the bed increases, which will accelerate the formation of coating layer, in turn, the stickiness of bed particles. At the same time, the agglomerate fraction almost increases linearly with the increase of fuel feeding rate. This indicates that the fuel feeding rate have an important effect on the agglomerate formation. On the one hand, the increase of fuel feeding rate accelerates the formation of coating layer. According to the coating-induced agglomeration mechanism the agglomerate fraction will increase. On the other hand, according to the melt-induced agglomeration mechanism, agglomerates are easily formed near the burning fuel particles because the high temperature causes more melt formation. The number of burning fuel particles will increases with the increase of fuel feeding rate. So the agglomerate fraction will increase too. It should be noted that the combustion stoichiometric factor change from 7.3 to 2.4 with the increase of fuel feeding from 10 g/h to 30 g/h when fluidization
H. Liu et al. / Bioresource Technology 100 (2009) 6505–6513
velocity is 0.1 m/s. Lin et al. (2003) reported that the stoichiometric factor has little influence on the defluidization time. Skrifvars
6509
et al. (1993) reported that reducing conditions have little influence on the sintering temperature of coal and biomass ashes.
Fig. 4. SEM micrographs of a bed particle.
6510
H. Liu et al. / Bioresource Technology 100 (2009) 6505–6513
So, in present study we also think that the change of defluidization time with fuel feeding rate only results from the effect of
burning fuel particles and take no account of the influence of stoichiometric factor.
Fig. 5. SEM micrographs of surfaces and cross sections of agglomerates.
6511
H. Liu et al. / Bioresource Technology 100 (2009) 6505–6513
Fig. 2c shows the effect of fluidization velocity on defluidization time and agglomerate fraction when fuel feeding rate is 15 g/h and bed temperature is 800 °C. Defluidization time increases with the increase of fluidization velocity. With the increase of fluidization velocity, the kinetic energy of bed particle increases. So, when collisions between bed particles take place, the impact force can more easily overcome the viscous force between them caused by their coating layers. Consequently, fluidization can maintain more time. At the same time, the agglomerate fraction decreases with the increase of fluidization velocity firstly and then closes to a constant. This indicates that increasing fluidization velocity can prevent the formation of agglomerates to some extent, but this effect is limited. According to above analyses, we can see that the stickiness of bed particles as a function of bed temperatures is the direct reason for bed defluidization. The formation of agglomerate seems to be controlled by the combination of bed temperature and fuel feeding rate. 3.2. Behaviors of alkali species in the rice straw ash It has been verified that the alkali species which cause the bed agglomeration come from the biomass ash. So, study on the behaviors of alkali species in rice straw ash is crucial for understanding the formation mechanisms of agglomerates and coating layers. Ash sample is obtained by combustion of rice straw particles in the muffle furnace at 850 °C for 10 h. Fig. 3A shows a close-up view of ash sample. The ash block can maintain a fixed shape and presents a little compression strength due to the sintering. As shown in Fig. 3B, ash particles are in the shape of strip and bonded by the some molten materials. Skrifvars et al. (1998) characterized the sintering tendency of ten biomass ashes and classified the ash components into three groups: simple alkali salts, silicates and the rest non-melt. Fig. 3C shows an amplified micrograph of the ash. We can see that the fusible compounds melt and coat the surface of ash particles during combustion (such as b and c). The EDS spot analyses indicate that these melts are rich in K or Na (points b and c in Table 2). For example, the content of K in melt b is 18.88%, and the content of Na in melt c is 9.84%. According to the experimental data and plenty of study results by other investigators (Skrifvars et al., 1998), these melts should be the alkali silicates. Based on the K2O–CaO–SiO2 phase diagram and the Na2O–SiO2 and K2O–SiO2 phase systems (Jenkins et al. (1995)) and the composition of these melts listed in Table 2 (normalized to the corresponding species including in the phase diagram), the melting point of these alkali silicates should be in the range 800–1200 °C. Fig. 3D shows the micrograph of the outer layer of a typical ash particle. This outer layer is composed of the infusible granulated materials arraying one by one. The EDS spot analysis shows they are the silicon dioxide grains (point d in Table 2). At the same time, these silicon dioxide grains are coated by the molten alkali silicates which can freely flow through the gaps between the silicon dioxide grains (points e, g and h in Fig. 3E). The EDS spot analysis on the infusible materials inside ash particle (point g in Table 2) shows
Table 3 EDS spot analyses of the coating layer. Points
a b c d e f g
they contains mainly Si and a little K, so there are not obvious melts on it. Form above findings we see that the alkali metals such as K and Na are mainly found in the outer layer of rice straw ash. The alkali silicates formed during combustion can melt and coat the surface of ash particles at high temperature. Further, it can be imagined that the sizes of ash particles must vary in a wider range due to the fragmentation and attrition in the bed. When the small-sized ash particles collide with bed particles, the molten alkali species on them will be transferred to surfaces of bed particles. In addition, the large-sized ash particles may increase the amount of melts on local surface of bed particle and may act as the necks for the agglomerate formation. So, the large- and small-sized ash particles may play the different roles in the bed agglomeration. 3.3. Formation mechanism of coating layer In previous studies, by means of SEM/EDS analyses on the cross sections of bed particles, the coating layer has been found to mainly contain the alkali silicates produced by the reaction between the alkali species from ash and the silica in the bed particles (Öhman et al., 2005; Nuutinen et al., 2004). However, the way by which alkali species are transferred onto the surface of bed particles has not been determined yet. In present study, we focus on the role of ash particles in the formation of coating layer. The SEM/EDS analyses are performed on the surfaces of bed particle samples. A typical bed particle collected after the test at bed temperature of 850 °C is analyzed here (see Fig. 4A). The EDS spot analyses are reported in Table 3. Fig. 4B and C shows two SEM micrographs of local surface of the bed particle. The large-sized ash particles in the molten state (>10 lm, indicted by arrow) can be found on the surface of bed particle. These large-sized ash particles usually contain K, Na, S and P (Such as point b in Table 3). Obviously, these large-sized ash particles should adhere to the bed particles by collisions. Fig. 4D, E, F, and G shows the amplified images of the local coating layer. The coating layer is composed of the melts and some infusible fine ash particles embedded in the melts. Elemental analyses show that these melts contain K, Na, and S (point a in Table 3) and the infusible ash particle mainly contain Ca and Mg (point g in Table 3). As shown in Fig. 4F, the coating layer behaves like viscous liquid and can flow along the surface of bed particle. And as shown in Fig. 4G, the coating layer is layered in some places. These findings indicate that the coating layer has been molten completely at bed temperature of 850 °C. So, the coating layer should mainly contain the alkali silicates with a lower melting point. More importantly, some viscous droplets in small size (<10 lm) are found on the surface of bed particle (point c in Fig. 4E). Elemental analyses show that these droplets are rich in Na and K (points c and d in Table 3). And they usually have the infusible ash kernels. So, they should be the small-sized ash particles deposited on the surface of bed particles. At the same time, the number density of them is so great that they can distribute on the bed particle surface densely and evenly. Moreover, it is obvious that these droplets are in the different sizes and at the different stages of melting. With the melting proceeding, the larger droplets disappear finally and enter into the coating layer, leaving the infusible ash kernels. In fact, the burning fuel
Elements (wt.%) Si
Na
K
Ca
Mg
O
S
P
44.69 46.66 40.65 44.16 52.65 47.99 50.55
4.52 1.75 5.09 3.45 1.63 0.29 0.12
3.48 2.38 4.53 3.74 2.39 0.47 0.24
0.87 0.92 0.46 0.76 0.75 0.23 1.16
1.02 0.71 2.54 0.26 0.07 0.23 1.32
40.89 45.47 44.51 45.58 39.86 47.99 38.78
1.21 0.68 0.75 0.59 0.93 0.81 0.99
3.32 1.42 1.47 1.47 1.73 1.84 6.84
Table 4 EDS spot analyses of necks on the surface of agglomerates. Points
a b
Elements (wt.%) Si
Na
K
Ca
Mg
O
C
30.46 59.68
1.63 3.35
3.95 9.55
1.17 1.78
0.72 1.25
14.45 24.39
47.62 0
6512
H. Liu et al. / Bioresource Technology 100 (2009) 6505–6513
particles are subject to strong fragmentation and attrition in the fluidized bed, so substantive small ash particles may be formed during combustion. These small-sized ash particles can be attached onto the sticky surfaces of bed particles easily. According to above findings we see that the small-sized ash particles may act as the carriers of the alkali species for the formation of coating layer. At the same time, the large number density and liquidity of them may be used to explain why coating layer can coat bed particles completely, even gaps and concaves on bed particles. 3.4. Formation mechanism of agglomerate
Ca Na Cl S P K
0 80 60 40 20 200 15 10 5 0 -5 40 30 20 10
O
P Si K
10
Si
Elemental composition (wt %)
Cl S
15 10 5 0 12 9 6 3 0 30 20 10 0 15 10 5 0 20
O
Elemental composition (wt %)
Na
Ca
In order to study the role of ash particles in the formation of agglomerates, the SEM/EDS analyses are performed on the surfaces and cross sections of agglomerates, respectively. The agglomerate samples are obtained from the same test used in last section. Here, the common characteristics of them are discussed. Fig. 5A shows the close-up view of a typical agglomerate. This agglomerate is about 3 mm in size and exhibits ellipse shape. It is composed of bed particles which should be partially bonded by the necks instead of embedded in the molten materials totally (see Fig. 5B). As shown in Fig. 5C, the molten ash particle (about 20 lm in size) also can be found on the surface of agglomerate. Fig. 5D, E and F shows three SEM micrographs of the typical necks found on the surface of agglomerates. In Fig. 5D, the neck adheres to the bed particles tightly through the molten materials. The EDS spot analysis can not be performed on it because it is located in concave. But this neck should be the outer layer of an ash particle identified by its shape. According to the findings in Section 3.2, this outer layer should be coated by the molten alkali salts. Here, the alkali salts react with the silicon in bed particles and produce the much stickier alkali silicates. In Fig. 5E, there exist some holes and bubbles in different sizes on the surface of the neck. The EDS spot analysis (point a in Table 4) shows that this neck contains not only K and Na but also a great amount of C. The only possible
explanation is that this neck comes from an ash particle in which carbon has not burned out completely. With combustion proceeding, gaseous production will release out of the neck, so the gas holes and bubbles are formed on the surface of it. In Fig. 5F, the neck looks compacter and the elemental analysis (point b in Table 4) shows that it also contain K and Na. Although carbon is not found in it, the small gas holes and combustion residues still can be found on its surface. These findings indicate that this neck should also come from an ash particle in which the carbon has burn out completely. Two typical illustrations of the cross sections of bed agglomerates are shown in Fig. 5G and I. The original images were digitally smoothed. The epoxy background appears dark and the sand particles appear bright in the images. Fig. 5G and I indicates that bed particles inside agglomerates are also partially bonded by the necks like that on the surface. The EDS line scans are utilized to characterize the necks inside agglomerates. Two hundred spot analyses are performed along each scan line (L1 and L2) and the results are shown in Fig. 6. Not only the variations of elemental composition but also the thicknesses of necks can be obtained by means of the EDS line scan. Two kinds of necks are found inside the agglomerates. For example, the thickness of neck shown in Fig. 5H is about 30 lm and the neck is rich in K, Na, S and Cl (see Fig. 6a). This kind neck should be composed of sulfates, chlorides and silicates of potassium and sodium. This elemental composition is close to the fuel ash, in other word, it should come from an ash particle. It accords with the melt-induced agglomeration mechanism. Another typical neck shown in Fig. 5J is about 10 lm in thickness and is much smaller than that in Fig. 5H. The EDS line scan shows that it also contains Na and K but is poor in Cl and S (see Fig. 6b). This indicts that the alkali silicates should be the dominate compound of this kind neck. This elemental composition is close to the coating layer. It accords with the coating-induced agglomeration mechanism. However, in present study, it is difficult to decide which kind of neck is the direct reason for agglomerate
0
10
20
30
40
50
60
70
80
15 10 5 0 12 9 6 3 0 30 20 10 0 15 10 5 0 25 20 15 10 5 0 80 60 40 20 0 15 10 5 0 40 30 20 10 0 0
10
20
Distance (µ m)
(a) L1 Fig. 6. EDS line scans on the necks inside agglomerates.
30
40
50
Distance (µ m)
(b) L2
60
70
80
H. Liu et al. / Bioresource Technology 100 (2009) 6505–6513
formation. But a conclusion can be drawn according to the above findings that the large-sized ash particles (>10 lm) can act as the necks in the formation of agglomerates. 4. Conclusion In this paper, the effects of fluidization velocity, bed temperature and fuel feeding rate on defluidization time and agglomerate fraction are studied. Necks in agglomerates and coating layers on bed particles are studied by means of SEM/DES analysis. Results show that the stickiness of coating layers is the direct reason for bed defluidization. The formation of agglomerate seems to be controlled by the combination of bed temperature and fuel feeding rate. Large-sized ash particles (>10 lm) may act as the necks in the formation of agglomerates. Melting of small-sized ash particles (<10 lm) will result in the formation of coating layer. Acknowledgement Authors thank the Harbin Institute of Technology (HIT) for the financial support on this postdoctoral investigation. References Arvelakis, S., Vourliotis, P., Kakaras, E., Koukios, E.G., 2001. Effect of leaching on the ash behavior of wheat straw and olive residue during fluidized bed combustion. Biomass and Bioenergy 20, 457–470. Bakker, R.R., Jenkins, B.M., Williams, R.B., 2002. Fluidized bed combustion of leached rice straw. Energy and Fuels 16, 356–365. Bartels, M., Lin, W.G., Nijenhuis, J., Kapteijn, F., Van Ommen, J.R., 2008. Agglomeration in fluidized beds at high temperature: mechanisms, detection and prevention. Progress in Energy and Combustion Science 34, 633–666. Brus, E., Ohman, M., Nordin, A., 2005. Mechanisms of bed agglomeration during fluidized-bed combustion of biomass fuels. Energy and Fuels 19, 825–832. Chirone, R., Miccio, F., Scala, F., 2006. Mechanism and prediction of bed agglomeration during fluidized bed combustion of a biomass fuel: effect of the scale. Chemical Engineering Journal 123, 71–80.
6513
Jenkins, B.M., Bakker, R.R., Wei, J.B., 1995. On the properties of washed straw. Biomass and Bioenergy 10, 177–200. Li, Z.Q., Zhao, W., Zhao, G.B., 2008. Effect of corn stalk length on combustion characteristics in a fixed bed. Energy and Fuels 22, 2009–2014. Lin, W.G., Johansen, K.D., Frandsen, F., 2003. Agglomeration in bio-fuel fires fluidized bed combustors. Chemical Engineering Journal 96, 171–185. Lind, T., Valmari, T., kauppinen, E., Nilsson, K., Sfiris, G., Maenhaut, W., 2000. Ash formation mechanism during combustion of wood in circulating fluidized beds. Proceedings of Combustion Institute 28, 2287–2295. Natarajan, E., Öhman, M., Gabra, M., Nordin, A., Liliedahl, T., Rao, A.N., 1998. Experimental determination of bed agglomeration tendencies of some common agricultural residues in fluidized bed combustion and gasification. Biomass and Bioenergy 15, 163–169. Nijenhuis, J., Korbee, R., Lensselink, J., 2007. A method for agglomeration detection and control in full-scale biomass fired fluidized. Chemical Engineering Science 62, 644–654. Nuutinen, L.H., Tiainen, M.S., Virtanen, M.E., Enestam, S.H., Laitinen, R.S., 2004. Coating layers on bed particles during biomass fuel combustion in fluidized-bed boiler. Energy and Fuels 18, 127–139. Öhman, M., Pommer, L., Nordin, A., 2005. Bed agglomeration characteristics and mechanisms during gasification and combustion of biomass fuels. Energy and Fuels 19, 1742–1748. Öhman, M., Nordin, A., Skrifvars, B.J., 2000. Bed agglomeration characteristics during fluidized bed combustion of biomass fuels. Energy and Fuels 14, 169– 178. Scala, F., Chirone, R., 2008. An SEM/EDS study of bed agglomerates formed during fluidized bed combustion of there biomass fuels. Biomass and Bioenergy 32, 252–266. Scala, F., Chirone, R., 2006. Characterization and detection of bed agglomeration during the fluidized bed of olive Husk. Energy and Fuels 20, 120–132. Skrifvars, B.J, Hupa, M., Patricainen. T., Laitinen, R., 1993. The effect of reducing conditions on coal ash sintering. In: Proceedings of the Engineering Foundation Conference: The Impact of Ash Deposition on Coal-Fired Plants. Skrifvars, B.J., Backman, R., Hupa, M., 1998. Characterization of the sintering tendency of ten biomass ashes in FBC conditions by a laboratory test and by phase equilibrium calculations. Fuel Processing Technology 56, 55–67. Vamvuka, D., Zografos, D., Alevizos, G., 2008. Control methods for mitigating biomass ash-related problems in fluidized beds. Bioresource Technology 99, 3534–3544. Werkelin, J., Skrifvars, B.j., Hupa, M., 2005. Ash-forming elements in four Scandinavian wood species. Part 1: summer harvest. Biomass and Bioenergy 29, 451–466.
Contents lists available at ScienceDirect
Bioresource Technology journal homepage: www.elsevier.com/locate/biortech
The role of ash particles in the bed agglomeration during the fluidized bed combustion of rice straw Huanpeng Liu a,b,*, Yujie Feng b, Shaohua Wu a, Dunyu Liu a a b
School of Energy Science and Engineering, Harbin Institute of Technology, Harbin, China School of Municipal and Environmental Engineering, Harbin Institute of Technology, Harbin, China
a r t i c l e
i n f o
Article history: Received 21 March 2009 Received in revised form 24 June 2009 Accepted 25 June 2009 Available online 6 August 2009 Keywords: Fluidized bed Defluidization time Agglomerate fraction Coating layer Neck
a b s t r a c t In this paper, the effects of fluidization velocity, bed temperature and fuel feeding rate on the defluidization time and agglomerate fraction in the fluidized bed combustion of rice straw were studied. The fuel ash, necks in agglomerates and coating layers of bed particles were studied by means of the scanning electron microscope, coupled with energy-dispersive spectroscopy (SEM/EDS). Results showed that the stickiness of bed particles induced by coating layers is the direct reason for bed defluidization. The alkali metals such as K and Na mainly exist in the outer layer of rice straw particles. During combustion the high temperature can cause the alkali species melting and coating the surfaces of ash particles. Consequently, ash particles become sticky and tend to adhere to the surfaces of bed particles. The large-sized ash particles may act as the necks in the formation of agglomerates. The small-sized ash particles may contribute to the formation of coating layers. Ó 2009 Elsevier Ltd. All rights reserved.
1. Introduction Biomass as a kind CO2-neutral energy source has been paid a great deal of attention (Li et al., 2008). At the same time, fluidized beds are widely used for the combustion of biomass because they are suitable for the low-grade fuels. However, in the practical operations, the bed agglomeration which can result in the unscheduled shutdown of fluidized bed reactors has become a serious problem. So, extensive studies have been carried out on it. It has been verified that the bed agglomeration is attributed to the alkali species in the biomass ash. The detection and counteraction methods have been studied (Nijenhuis et al., 2007; Vamvuka et al., 2008; Arvelakis et al., 2001; Bartels et al., 2008; Bakker et al., 2002). The effects of running parameters (such as bed temperature, fluidization velocity, etc.) on the bed agglomeration were studied by Lin et al. (2003). The ash-forming elements in the different biomass were analyzed and the ash formation mechanisms during the combustion of biomass in fluidized beds were studied too (Lind et al., 2000; Werkelin et al., 2005). It has been widely accepted that the alkali species in the biomass ash can be transferred to the surface of bed particles and result in the formation of coating layer during combustion. The coating layer consists of several superim* Corresponding author. Address: School of Energy Science and Engineering, No. 457 mailbox, Harbin Institute of Technology, Harbin 150001, China. Tel.: +86 451 86412578; fax: +86 451 86412528. E-mail address: [email protected] (H. Liu). 0960-8524/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.biortech.2009.06.098
posed layers. The innermost layer mainly contains alkali silicates, whereas the outermost layer is rich in calcium or magnesium (Nuutinen et al., 2004). Although the deposition of fine ash particles, condensation of volatile alkali species and the collisions between the bed particles and burning char particles have been reported as critical transport mechanisms (Scala and Chirone, 2006, 2008), the formation process of the coating layer has not been determined up to now. Meanwhile, studies have been conducted on the bed agglomeration mechanism. Up until now, two different types of agglomeration formation have been proposed: melt-induced agglomeration and coating-induced agglomeration. In the melt-induced agglomeration, the molten ash particles are considered to act as the necks to bond the bed particles in the agglomerate formation process (Chirone et al., 2006; Lin et al., 2003). In the coating-induced agglomeration, the melting of coating layers of bed particles is considered to be responsible for the bed agglomeration (Natarajan et al., 1998; Öhman et al., 2000, 2005; Brus et al., 2005). But the formation mechanism of the agglomerate has not been determined because there are not enough experimental evidences to verify these hypotheses. In present work, rice straw is used as the study object. The experiment is carried out in a lab-scale bubbling fluidized bed combustor. The effects of fluidization velocity, bed temperature and fuel feeing rate on the defluidization time and agglomerate fraction are studied. As for the bed agglomeration mechanism, we focus on the formation processes of agglomerates and coating layers, specially the role of ash particles in these formation
6506
H. Liu et al. / Bioresource Technology 100 (2009) 6505–6513
1. pressure meter 2. temperature controllor 3. thermalcouple 4. heating wire 5. refractory protection 6. heat insulator 7. distributor 8. flow-meter 9. Compressor 10.windbox 11.discharge pipe Fig. 1. Schematic of the experimental apparatus.
(a)
(b)
6
200
V=0.1m/s
100
F=15g/h
3
t f
80
2
60 40
1
Defluidization time t (min)
4
120
20 0 740
760
780
800
820
840
860
880
900
0 920
T=800°C
t
V=0.1m/s
f
4
125 100
3
75
2
50 1
25 0
0 10
Bed temperatureT (°C)
(c)
20
25
30
6
t
T=800°C
f F=15g/h
5
150 4
125
3
100 75
2
50 1
25
0.07
0.08
0.09
0.10
0.11
0.12
Agglomerates fraction f (%)
175
Defluidization time t (min)
15
Fuel feeding rateF (g/h)
200
0 0.06
5
150
Agglomerate fraction f (%)
140
6
200 175
5
160
Agglomerates fraction f (%)
Defluidization time t (min)
180
0 0.13
Fluidization velocity V (m/s) Fig. 2. Effects of running parameters on the defluidization time and agglomerate fraction.
processes. The SEM/EDS analysis was carried out on the outer surfaces of ash particles, the outer surfaces and cross sections of
agglomerates, as well as the outer surfaces of bed particles. Some valuable findings are obtained.
H. Liu et al. / Bioresource Technology 100 (2009) 6505–6513
2. Methods The experimental apparatus is schematically shown in Fig. 1. The reactor is made of high temperature resistant steel tube with an inner diameter of 32 mm and a total height of 600 mm. A perforated steel plate distributor is located at the bottom of bed. The reactor is heated by means of heating wire
6507
with a standard power of 2.6 kW. Quartz sand sieved to a size range 200–300 lm is used as the bed material. The sand contains >98% SiO2 and the initial mass of bed material of each test is 200 g. The bed temperature is continuously monitored and controlled by the thermocouples. Air is fed to the windbox by an air compressor. The fluidization velocity is regulated by a flow-meter.
Fig. 3. Close-up view and SEM micrographs of rice straw ash.
6508
H. Liu et al. / Bioresource Technology 100 (2009) 6505–6513
Table 1 Elemental analysis of rice straw and ash used in experiments. Moisture
Ash
Volatile
Fixed carbon
Heat value
Proximate analysis of rice straw (as received) (wt.%) 4.99 13.87
65.05
16.09
13,860 kJ/kg
Carbon
Nitrogen
Sulphur
Oxygen
1.09
0.08
36.66
Hydrogen
Ultimate analysis of rice straw (dry basis) (wt.%) 41.93 5.64 Na2O
MgO
Al2O3
Ultimate analysis of rice straw ash (wt.%) 1.98 2.47 1.94
SiO2
P2O5
SO3
K2O
CaO
Fe2O3
56.23
1.93
4.95
19.4
10.12
0.98
Rice straw used in the experiment comes from a farm in Heilongjiang province in China. Table 1 shows the proximate and ultimate analyses of rice straw and the composition of rice straw ash. Rice straw is broken to the size range 1–2 mm by a biomass crushing engine before experiments. Fuel particles are store in a container and continuously fed into the combustor by a screw feeder. Bed temperature will increase as soon as the fuel is introduced. In order to keep bed temperature at a given value, a trial test is performed before every case to determine the pre-heated temperature of bed before feeding fuel. Even so, bed temperature still fluctuates slightly around the given value. So, bed temperature is defined as the averaged value throughout the test. In present study, fluidization velocity varies from 0.067 to 1.2 m/s and fuel feeding rate varies from 10 to 30 g/h. The excess air factors for all tests are in the range 1.5–8.0. So, all tests should run under the combustion condition. The combustor start-up is accomplished by electrically heating the bed of fresh inert sand. When the bed temperature reaches the desired value, the test and fuel feeding begin. During the test, stir will be performed as soon as the bed defluidization occurs. The test is considered to be over when the bed fluidization can not be recovered by the stir. When test ends, bed materials is discharged and sieved. In this study, agglomerate fraction is defined as the weight ratio of agglomerates to the bed material. It is used to indicate the extent of agglomerate formation among tests with different running parameters. Here, the sizes of agglomerates are larger than 2 mm. Because agglomerates are very fragile, the value of agglomerate fraction is sensitive to the way of handing and collection. So, every test is repeated for three times. The agglomerate fraction under one running condition is the average of the three repeated tests. The defluidization time is defined as the time between the start of fuel feeding to the defluidization of the bed. In order to study the cross sections of the agglomerates, some samples are mounted in epoxy resin, cut, polished with sandpapers, finally, coated with gold layer for the SEM/DES analyses.
Table 2 EDS spot analyses of the rice straw ash. Points
a b c d e f g h
Elements (wt.%) Si
Na
K
Ca
Mg
O
Cl
S
P
44.97 42.61 55.87 51.92 44.87 39.79 51.41 52.06
0.97 0.73 9.84 1.00 5.01 3.17 0.49 5.49
3.47 18.88 0.97 0.67 8.60 8.16 1.3 10.84
4.55 19.12 0.64 0.0 5.23 5.0 1.0 0.0
0.18 0.58 0.0 0.42 0.62 0.78 0.35 1.05
37.54 15.54 28.79 43.33 29.93 39.76 44.24 23.22
0.47 1.76 0.0 0.32 1.99 1.21 0.0 0.0
1.04 0.0 1.48 0.72 1.18 0.75 0.18 0.44
6.81 0.77 2.41 1.64 2.58 1.38 1.03 6.79
3. Results and discussion 3.1. Effects of running parameters on the defluidization time and agglomerate fraction Fig. 2a shows the effect of bed temperature (T) on defluidization time (t) and agglomerate fraction (f) when fuel feeding rate (F) is 15 g/h and fluidization velocity (V) is 0.1 m/s. There is no obvious tendency of bed defluidization when bed temperature is below 750 °C. When bed temperature is above 750 °C, bed particles appear sticky. It has been known that the stickiness of bed particles is attributed to the melting of low-melting point alkali silicates coating layer on the surface of bed particles. Defluidization time decreases with the bed temperature increasing from 750 to 900 °C. This indicates that some compounds in the coating layer begin to melt when bed temperature is 750 °C. As bed temperature increases, the fraction of melts in the coating layers of bed particles increases. Then the viscosity of coating layer increases. The viscous forces between the bed particles increase too. So, the defluidization time will decrease with the increase of bed temperature. At the same time, agglomerate fraction increases firstly and reaches a maximum value at the bed temperature of 850 °C. Interestingly, when bed temperature increases as high as 900 °C, although bed defluidization occurs soon, agglomerate fraction decreases a little instead of increases. This indicates that the stickiness of bed particle induced by the melting of coating layers is the direct reason for the bed defluidizaton, but is not the only reason for the agglomerate formation. Fig. 2b shows the effect of fuel feeding rate on defluidization time and agglomerate fraction when bed temperature is 800 °C and fluidization velocity is 0.1 m/s. Defluidization time decreases with the increase of fuel feeding rate. It has been accepted that the alkali species which cause the formation of coating layers come form the biomass ash. So, with fuel feeding rate increasing, the amount of alkali species in the bed increases, which will accelerate the formation of coating layer, in turn, the stickiness of bed particles. At the same time, the agglomerate fraction almost increases linearly with the increase of fuel feeding rate. This indicates that the fuel feeding rate have an important effect on the agglomerate formation. On the one hand, the increase of fuel feeding rate accelerates the formation of coating layer. According to the coating-induced agglomeration mechanism the agglomerate fraction will increase. On the other hand, according to the melt-induced agglomeration mechanism, agglomerates are easily formed near the burning fuel particles because the high temperature causes more melt formation. The number of burning fuel particles will increases with the increase of fuel feeding rate. So the agglomerate fraction will increase too. It should be noted that the combustion stoichiometric factor change from 7.3 to 2.4 with the increase of fuel feeding from 10 g/h to 30 g/h when fluidization
H. Liu et al. / Bioresource Technology 100 (2009) 6505–6513
velocity is 0.1 m/s. Lin et al. (2003) reported that the stoichiometric factor has little influence on the defluidization time. Skrifvars
6509
et al. (1993) reported that reducing conditions have little influence on the sintering temperature of coal and biomass ashes.
Fig. 4. SEM micrographs of a bed particle.
6510
H. Liu et al. / Bioresource Technology 100 (2009) 6505–6513
So, in present study we also think that the change of defluidization time with fuel feeding rate only results from the effect of
burning fuel particles and take no account of the influence of stoichiometric factor.
Fig. 5. SEM micrographs of surfaces and cross sections of agglomerates.
6511
H. Liu et al. / Bioresource Technology 100 (2009) 6505–6513
Fig. 2c shows the effect of fluidization velocity on defluidization time and agglomerate fraction when fuel feeding rate is 15 g/h and bed temperature is 800 °C. Defluidization time increases with the increase of fluidization velocity. With the increase of fluidization velocity, the kinetic energy of bed particle increases. So, when collisions between bed particles take place, the impact force can more easily overcome the viscous force between them caused by their coating layers. Consequently, fluidization can maintain more time. At the same time, the agglomerate fraction decreases with the increase of fluidization velocity firstly and then closes to a constant. This indicates that increasing fluidization velocity can prevent the formation of agglomerates to some extent, but this effect is limited. According to above analyses, we can see that the stickiness of bed particles as a function of bed temperatures is the direct reason for bed defluidization. The formation of agglomerate seems to be controlled by the combination of bed temperature and fuel feeding rate. 3.2. Behaviors of alkali species in the rice straw ash It has been verified that the alkali species which cause the bed agglomeration come from the biomass ash. So, study on the behaviors of alkali species in rice straw ash is crucial for understanding the formation mechanisms of agglomerates and coating layers. Ash sample is obtained by combustion of rice straw particles in the muffle furnace at 850 °C for 10 h. Fig. 3A shows a close-up view of ash sample. The ash block can maintain a fixed shape and presents a little compression strength due to the sintering. As shown in Fig. 3B, ash particles are in the shape of strip and bonded by the some molten materials. Skrifvars et al. (1998) characterized the sintering tendency of ten biomass ashes and classified the ash components into three groups: simple alkali salts, silicates and the rest non-melt. Fig. 3C shows an amplified micrograph of the ash. We can see that the fusible compounds melt and coat the surface of ash particles during combustion (such as b and c). The EDS spot analyses indicate that these melts are rich in K or Na (points b and c in Table 2). For example, the content of K in melt b is 18.88%, and the content of Na in melt c is 9.84%. According to the experimental data and plenty of study results by other investigators (Skrifvars et al., 1998), these melts should be the alkali silicates. Based on the K2O–CaO–SiO2 phase diagram and the Na2O–SiO2 and K2O–SiO2 phase systems (Jenkins et al. (1995)) and the composition of these melts listed in Table 2 (normalized to the corresponding species including in the phase diagram), the melting point of these alkali silicates should be in the range 800–1200 °C. Fig. 3D shows the micrograph of the outer layer of a typical ash particle. This outer layer is composed of the infusible granulated materials arraying one by one. The EDS spot analysis shows they are the silicon dioxide grains (point d in Table 2). At the same time, these silicon dioxide grains are coated by the molten alkali silicates which can freely flow through the gaps between the silicon dioxide grains (points e, g and h in Fig. 3E). The EDS spot analysis on the infusible materials inside ash particle (point g in Table 2) shows
Table 3 EDS spot analyses of the coating layer. Points
a b c d e f g
they contains mainly Si and a little K, so there are not obvious melts on it. Form above findings we see that the alkali metals such as K and Na are mainly found in the outer layer of rice straw ash. The alkali silicates formed during combustion can melt and coat the surface of ash particles at high temperature. Further, it can be imagined that the sizes of ash particles must vary in a wider range due to the fragmentation and attrition in the bed. When the small-sized ash particles collide with bed particles, the molten alkali species on them will be transferred to surfaces of bed particles. In addition, the large-sized ash particles may increase the amount of melts on local surface of bed particle and may act as the necks for the agglomerate formation. So, the large- and small-sized ash particles may play the different roles in the bed agglomeration. 3.3. Formation mechanism of coating layer In previous studies, by means of SEM/EDS analyses on the cross sections of bed particles, the coating layer has been found to mainly contain the alkali silicates produced by the reaction between the alkali species from ash and the silica in the bed particles (Öhman et al., 2005; Nuutinen et al., 2004). However, the way by which alkali species are transferred onto the surface of bed particles has not been determined yet. In present study, we focus on the role of ash particles in the formation of coating layer. The SEM/EDS analyses are performed on the surfaces of bed particle samples. A typical bed particle collected after the test at bed temperature of 850 °C is analyzed here (see Fig. 4A). The EDS spot analyses are reported in Table 3. Fig. 4B and C shows two SEM micrographs of local surface of the bed particle. The large-sized ash particles in the molten state (>10 lm, indicted by arrow) can be found on the surface of bed particle. These large-sized ash particles usually contain K, Na, S and P (Such as point b in Table 3). Obviously, these large-sized ash particles should adhere to the bed particles by collisions. Fig. 4D, E, F, and G shows the amplified images of the local coating layer. The coating layer is composed of the melts and some infusible fine ash particles embedded in the melts. Elemental analyses show that these melts contain K, Na, and S (point a in Table 3) and the infusible ash particle mainly contain Ca and Mg (point g in Table 3). As shown in Fig. 4F, the coating layer behaves like viscous liquid and can flow along the surface of bed particle. And as shown in Fig. 4G, the coating layer is layered in some places. These findings indicate that the coating layer has been molten completely at bed temperature of 850 °C. So, the coating layer should mainly contain the alkali silicates with a lower melting point. More importantly, some viscous droplets in small size (<10 lm) are found on the surface of bed particle (point c in Fig. 4E). Elemental analyses show that these droplets are rich in Na and K (points c and d in Table 3). And they usually have the infusible ash kernels. So, they should be the small-sized ash particles deposited on the surface of bed particles. At the same time, the number density of them is so great that they can distribute on the bed particle surface densely and evenly. Moreover, it is obvious that these droplets are in the different sizes and at the different stages of melting. With the melting proceeding, the larger droplets disappear finally and enter into the coating layer, leaving the infusible ash kernels. In fact, the burning fuel
Elements (wt.%) Si
Na
K
Ca
Mg
O
S
P
44.69 46.66 40.65 44.16 52.65 47.99 50.55
4.52 1.75 5.09 3.45 1.63 0.29 0.12
3.48 2.38 4.53 3.74 2.39 0.47 0.24
0.87 0.92 0.46 0.76 0.75 0.23 1.16
1.02 0.71 2.54 0.26 0.07 0.23 1.32
40.89 45.47 44.51 45.58 39.86 47.99 38.78
1.21 0.68 0.75 0.59 0.93 0.81 0.99
3.32 1.42 1.47 1.47 1.73 1.84 6.84
Table 4 EDS spot analyses of necks on the surface of agglomerates. Points
a b
Elements (wt.%) Si
Na
K
Ca
Mg
O
C
30.46 59.68
1.63 3.35
3.95 9.55
1.17 1.78
0.72 1.25
14.45 24.39
47.62 0
6512
H. Liu et al. / Bioresource Technology 100 (2009) 6505–6513
particles are subject to strong fragmentation and attrition in the fluidized bed, so substantive small ash particles may be formed during combustion. These small-sized ash particles can be attached onto the sticky surfaces of bed particles easily. According to above findings we see that the small-sized ash particles may act as the carriers of the alkali species for the formation of coating layer. At the same time, the large number density and liquidity of them may be used to explain why coating layer can coat bed particles completely, even gaps and concaves on bed particles. 3.4. Formation mechanism of agglomerate
Ca Na Cl S P K
0 80 60 40 20 200 15 10 5 0 -5 40 30 20 10
O
P Si K
10
Si
Elemental composition (wt %)
Cl S
15 10 5 0 12 9 6 3 0 30 20 10 0 15 10 5 0 20
O
Elemental composition (wt %)
Na
Ca
In order to study the role of ash particles in the formation of agglomerates, the SEM/EDS analyses are performed on the surfaces and cross sections of agglomerates, respectively. The agglomerate samples are obtained from the same test used in last section. Here, the common characteristics of them are discussed. Fig. 5A shows the close-up view of a typical agglomerate. This agglomerate is about 3 mm in size and exhibits ellipse shape. It is composed of bed particles which should be partially bonded by the necks instead of embedded in the molten materials totally (see Fig. 5B). As shown in Fig. 5C, the molten ash particle (about 20 lm in size) also can be found on the surface of agglomerate. Fig. 5D, E and F shows three SEM micrographs of the typical necks found on the surface of agglomerates. In Fig. 5D, the neck adheres to the bed particles tightly through the molten materials. The EDS spot analysis can not be performed on it because it is located in concave. But this neck should be the outer layer of an ash particle identified by its shape. According to the findings in Section 3.2, this outer layer should be coated by the molten alkali salts. Here, the alkali salts react with the silicon in bed particles and produce the much stickier alkali silicates. In Fig. 5E, there exist some holes and bubbles in different sizes on the surface of the neck. The EDS spot analysis (point a in Table 4) shows that this neck contains not only K and Na but also a great amount of C. The only possible
explanation is that this neck comes from an ash particle in which carbon has not burned out completely. With combustion proceeding, gaseous production will release out of the neck, so the gas holes and bubbles are formed on the surface of it. In Fig. 5F, the neck looks compacter and the elemental analysis (point b in Table 4) shows that it also contain K and Na. Although carbon is not found in it, the small gas holes and combustion residues still can be found on its surface. These findings indicate that this neck should also come from an ash particle in which the carbon has burn out completely. Two typical illustrations of the cross sections of bed agglomerates are shown in Fig. 5G and I. The original images were digitally smoothed. The epoxy background appears dark and the sand particles appear bright in the images. Fig. 5G and I indicates that bed particles inside agglomerates are also partially bonded by the necks like that on the surface. The EDS line scans are utilized to characterize the necks inside agglomerates. Two hundred spot analyses are performed along each scan line (L1 and L2) and the results are shown in Fig. 6. Not only the variations of elemental composition but also the thicknesses of necks can be obtained by means of the EDS line scan. Two kinds of necks are found inside the agglomerates. For example, the thickness of neck shown in Fig. 5H is about 30 lm and the neck is rich in K, Na, S and Cl (see Fig. 6a). This kind neck should be composed of sulfates, chlorides and silicates of potassium and sodium. This elemental composition is close to the fuel ash, in other word, it should come from an ash particle. It accords with the melt-induced agglomeration mechanism. Another typical neck shown in Fig. 5J is about 10 lm in thickness and is much smaller than that in Fig. 5H. The EDS line scan shows that it also contains Na and K but is poor in Cl and S (see Fig. 6b). This indicts that the alkali silicates should be the dominate compound of this kind neck. This elemental composition is close to the coating layer. It accords with the coating-induced agglomeration mechanism. However, in present study, it is difficult to decide which kind of neck is the direct reason for agglomerate
0
10
20
30
40
50
60
70
80
15 10 5 0 12 9 6 3 0 30 20 10 0 15 10 5 0 25 20 15 10 5 0 80 60 40 20 0 15 10 5 0 40 30 20 10 0 0
10
20
Distance (µ m)
(a) L1 Fig. 6. EDS line scans on the necks inside agglomerates.
30
40
50
Distance (µ m)
(b) L2
60
70
80
H. Liu et al. / Bioresource Technology 100 (2009) 6505–6513
formation. But a conclusion can be drawn according to the above findings that the large-sized ash particles (>10 lm) can act as the necks in the formation of agglomerates. 4. Conclusion In this paper, the effects of fluidization velocity, bed temperature and fuel feeding rate on defluidization time and agglomerate fraction are studied. Necks in agglomerates and coating layers on bed particles are studied by means of SEM/DES analysis. Results show that the stickiness of coating layers is the direct reason for bed defluidization. The formation of agglomerate seems to be controlled by the combination of bed temperature and fuel feeding rate. Large-sized ash particles (>10 lm) may act as the necks in the formation of agglomerates. Melting of small-sized ash particles (<10 lm) will result in the formation of coating layer. Acknowledgement Authors thank the Harbin Institute of Technology (HIT) for the financial support on this postdoctoral investigation. References Arvelakis, S., Vourliotis, P., Kakaras, E., Koukios, E.G., 2001. Effect of leaching on the ash behavior of wheat straw and olive residue during fluidized bed combustion. Biomass and Bioenergy 20, 457–470. Bakker, R.R., Jenkins, B.M., Williams, R.B., 2002. Fluidized bed combustion of leached rice straw. Energy and Fuels 16, 356–365. Bartels, M., Lin, W.G., Nijenhuis, J., Kapteijn, F., Van Ommen, J.R., 2008. Agglomeration in fluidized beds at high temperature: mechanisms, detection and prevention. Progress in Energy and Combustion Science 34, 633–666. Brus, E., Ohman, M., Nordin, A., 2005. Mechanisms of bed agglomeration during fluidized-bed combustion of biomass fuels. Energy and Fuels 19, 825–832. Chirone, R., Miccio, F., Scala, F., 2006. Mechanism and prediction of bed agglomeration during fluidized bed combustion of a biomass fuel: effect of the scale. Chemical Engineering Journal 123, 71–80.
6513
Jenkins, B.M., Bakker, R.R., Wei, J.B., 1995. On the properties of washed straw. Biomass and Bioenergy 10, 177–200. Li, Z.Q., Zhao, W., Zhao, G.B., 2008. Effect of corn stalk length on combustion characteristics in a fixed bed. Energy and Fuels 22, 2009–2014. Lin, W.G., Johansen, K.D., Frandsen, F., 2003. Agglomeration in bio-fuel fires fluidized bed combustors. Chemical Engineering Journal 96, 171–185. Lind, T., Valmari, T., kauppinen, E., Nilsson, K., Sfiris, G., Maenhaut, W., 2000. Ash formation mechanism during combustion of wood in circulating fluidized beds. Proceedings of Combustion Institute 28, 2287–2295. Natarajan, E., Öhman, M., Gabra, M., Nordin, A., Liliedahl, T., Rao, A.N., 1998. Experimental determination of bed agglomeration tendencies of some common agricultural residues in fluidized bed combustion and gasification. Biomass and Bioenergy 15, 163–169. Nijenhuis, J., Korbee, R., Lensselink, J., 2007. A method for agglomeration detection and control in full-scale biomass fired fluidized. Chemical Engineering Science 62, 644–654. Nuutinen, L.H., Tiainen, M.S., Virtanen, M.E., Enestam, S.H., Laitinen, R.S., 2004. Coating layers on bed particles during biomass fuel combustion in fluidized-bed boiler. Energy and Fuels 18, 127–139. Öhman, M., Pommer, L., Nordin, A., 2005. Bed agglomeration characteristics and mechanisms during gasification and combustion of biomass fuels. Energy and Fuels 19, 1742–1748. Öhman, M., Nordin, A., Skrifvars, B.J., 2000. Bed agglomeration characteristics during fluidized bed combustion of biomass fuels. Energy and Fuels 14, 169– 178. Scala, F., Chirone, R., 2008. An SEM/EDS study of bed agglomerates formed during fluidized bed combustion of there biomass fuels. Biomass and Bioenergy 32, 252–266. Scala, F., Chirone, R., 2006. Characterization and detection of bed agglomeration during the fluidized bed of olive Husk. Energy and Fuels 20, 120–132. Skrifvars, B.J, Hupa, M., Patricainen. T., Laitinen, R., 1993. The effect of reducing conditions on coal ash sintering. In: Proceedings of the Engineering Foundation Conference: The Impact of Ash Deposition on Coal-Fired Plants. Skrifvars, B.J., Backman, R., Hupa, M., 1998. Characterization of the sintering tendency of ten biomass ashes in FBC conditions by a laboratory test and by phase equilibrium calculations. Fuel Processing Technology 56, 55–67. Vamvuka, D., Zografos, D., Alevizos, G., 2008. Control methods for mitigating biomass ash-related problems in fluidized beds. Bioresource Technology 99, 3534–3544. Werkelin, J., Skrifvars, B.j., Hupa, M., 2005. Ash-forming elements in four Scandinavian wood species. Part 1: summer harvest. Biomass and Bioenergy 29, 451–466.