Ash Deposition on Heat Transfer Tubes during Combustion of

Mar 9, 2006 - The deposits from wood combustion or from wood contaminated .... WP100, 22, 0.2, 34, 1.1, 1.4, 3 ... a D50% is the diameter whose collec...
0 downloads 0 Views 3MB Size
Energy & Fuels 2006, 20, 1001-1007

1001

Ash Deposition on Heat Transfer Tubes during Combustion of Demolition Wood Lars-Erik Åmand,*,† Bo Leckner,† David Eskilsson,‡ and Claes Tullin‡ Department of Energy and EnVironment, DiVision of Energy ConVersion, Chalmers UniVersity of Technology, SE-412 96 Go¨teborg, Sweden, and SP Swedish National Testing and Research Institute, Department of Energy Technology, Postbox 857, SE-501 15 Borås, Sweden ReceiVed NoVember 23, 2005. ReVised Manuscript ReceiVed February 1, 2006

Wood from old buildings, demolition wood, has a considerable energy value and can be used as a fuel. However, the material contains impurities that may have harmful effects. This investigation, carried out in a fluidized-bed boiler, focuses on the formation of deposits on heat transfer surfaces during combustion of demolition wood. To avoid inhomogeneity and variations in composition of the fuel during the tests, highquality wood was used and two substances that are suspected to contribute most to deposits, chlorine and zinc (from paint), were added in predetermined quantities. It was found that chlorine was the most harmful constituent, promoting deposits in the form of potassium chloride and also zinc chloride. The deposits from wood combustion or from wood contaminated with zinc are deemed not to be excessively harmful with respect to corrosion as long as the chlorine concentration is low and oxidizing conditions prevail.

Introduction Demolition wood remains after reconstruction of buildings when recyclable parts of the material have been removed. Demolition wood has a high heating value and is useful as a fuel, but with the potential drawback of being contaminated in various ways. Enhanced deposits and corrosion were observed in boilers using demolition wood as an additional fuel, but the origin of the problems was not entirely clear. Therefore, a field study was carried out to measure deposits and corrosion in gratefired as well as in fluidized-bed boilers.1 This study identified zinc as an important component, together with K, Na, S, and Cl, leading to enhanced deposit formation during operation with demolition wood alone or together with other types of wood (wood waste from the forest). The results were slightly different in different boilers, so design and operational conditions are also important for deposit formation. Zinc is a common heavy metal in deposits on incinerator tubes during combustion of waste in general. Zinc chloride is known to have a particularly deleterious impact on tubes together with KCl, because it promotes low-temperature melts and, hence, corrosion.2 In the present work, attention is focused on the formation of enhanced deposits, deposits that could lead to corrosion and blockage of gas paths in boilers during combustion of demolition wood. In addition to potassium included in the wood ash, zinc oxide, originating from paint remaining on the wood, and an increased content of chlorine, worsened by plastics unintentionally mixed into the material, are the principal suspects of disturbances. * Corresponding author. Phone: +46 31 772 1439. Fax: +46 31 772 3592. E-mail: [email protected]. † Chalmers University of Technology. ‡ SP Swedish National Testing and Research Institute. (1) Andersson, C.; Ho¨gberg, J. Reasons for ash-related operation problems during combustion of demolition wood, Report 733 (Fg-821) (in Swedish with English summary), Va¨rmeforsk, ISSN 0282-3771, Stockholm, 2001. (2) Sarofim, A. F.; Helble, J. J. In The Impact of Ash Deposition on Coal Fired Plants; Williamson, J., Wigley, F., Eds.; Taylor & Francis: Washington, DC, 1993; pp 567-582.

Other contaminants, such as lead compounds, could be significant in some wood wastes, but zinc is a suitable representative of such materials, although their behaviors are not entirely identical. Hence, for the present purpose, potassium, zinc, and chlorine are the most suitable species to be included in an investigation, the purpose of which is to study the initial deposition of mineral fuel constituents on a probe, simulating a superheater tube in a boiler that is fired with demolition wood. The investigation especially considers the composition of the particle flow in the flue gas path and deposit formation on heat transfer tubes, represented by the probe. Under oxidizing conditions, zinc oxide is thermally stable in the temperature range usually encountered in bed combustion; however, under reducing conditions, which can occur locally in a furnace, ZnO may be reduced to metallic zinc according to

ZnO(s) + CO(g) S Zn(s) + CO2(g) Zn(s) S Zn(g)

(1)

In a reducing environment, metallic zinc is evaporated at 500-700 °C and then again oxidized if entering an oxidizing region in the furnace:

2Zn(g) + O2(g) S 2ZnO(s)

(2)

In case of severely reducing conditions in furnaces, oxidation does not take place, and metallic zinc may be deposited on tubes. ZnO can react with chlorine

ZnO(s) + 2HCl(g) S ZnCl2(s) + H2O(g) ZnCl2(s) S ZnCl2(g)

(3)

Zinc oxide can also react with sulfur to form ZnSO4, which can be evaporated in a combustion environment. Equilibrium

10.1021/ef0503919 CCC: $33.50 © 2006 American Chemical Society Published on Web 03/09/2006

1002 Energy & Fuels, Vol. 20, No. 3, 2006

Åmand et al. Table 2. Average Fuel Properties (daf means dry and ash-free)a

Table 1. Operation Data of the Boiler parameter

value

load, MWth bed temperature (bottom), °C bed temperature (top), °C temperature, after primary cyclone, °C temperature at particle sampling, °C temperature before bag filter, °C excess-air ratio primary/total air-flow, % total furnace pressure drop, kPa superficial velocity (top), m/s

6.3 840 870 825 270 150 1.2 60 6.4 5

wood pellets

calculations (for instance, ref 3) show that under oxidizing conditions the volatility increases with the concentration of chlorine as indicated by reaction 3; however, there are several other factors that may have an influence on the production of ZnCl2(g), for instance, the presence and concentration of water vapor (reaction 3), and also complex compounds, such as metal aluminates for example. This idea was further emphasized by Mininni et al.4 in explaining their result that zinc was not particularly volatile during sludge combustion in contrast to their thermodynamic estimates. Also, Lind et al.,5 investigating ash behavior in connection with combustion of forest wastes, found that most zinc was removed with the bottom ash and only 0.10.3% were in the gas phase at 830 °C. Most of the potassium (or sodium) in wood is gasified and eventually reacts in gas phase with available chlorine or sulfur. If sulfur is present, alkali (K or Na, denoted M) chloride, possibly formed, can be converted into a sulfate through the overall reaction

2MCl + SO2 + 1/2O2 + H2O S M2SO4 + 2HCl

(4)

Backman et al.6 showed by equilibrium calculations that during oxidizing conditions the presence of alkali (but not sulfur) can affect both the amount of zinc volatilization and the chemical form (ZnO or ZnCl2) in the flue gas. This effect arises because KCl and NaCl are more stable than ZnCl2. Adding sulfur to the calculations will increase the available chlorine as HCl (see Reaction 4) leading to formation of ZnCl2. Using the full element mechanism in the calculations (Backman et al.6) results only in a minor production of ZnCl2 due to formation of other complex compounds that are not volatile. Here, attention will be paid to the species mentioned, collected from the flue gases by a sampling probe, as well as on a solid surface (the deposition probe). Experimental Section The Boiler. The tests were carried out in the 12-MWth circulating fluidized-bed (CFB) boiler at Chalmers University of Technology. A description of the boiler is found, for instance, in MiettinenWestberg et al.7 Load and operation data employed, typical for a CFB boiler, are summarized in Table 1. These data were kept constant during the tests with high accuracy; the maximum deviation (3) Verhulst, D.; Buekens, A.; Spencer, P. J.; Eriksson, G. EnViron. Sci. Technol. 1996, 30, 50-56. (4) Mininni, G.; Marani, D.; Braguglia, C. M.; Guerriero, E.; Sbrilli, A. In Proceedings of the 17th International Conference on Fluidized Bed Combustion; American Society of Mechanical Engineers: New York, 2003, Paper FBC2003-105. (5) Lind, T.; Valmari, T.; Kauppinen, E. I.; Sfiris, G.; Nilsson, K.; Maenhaut, W. EnViron. Sci. Technol. 1999, 33, 496-502. (6) Backman, R.; Hupa, M.; Hiltunen, M.; Peltola, K. In Proceedings of the 18th International Conference on Fluidized Bed Combustion; American Society of Mechanical Engineers: New York, 2005, Paper FBC2005-78074. (7) Miettinen-Westberg, H.; Bystro¨m, M.; Leckner, B. Energy Fuels 2003, 17, 18-28.

proximate analysis moisture (mass %, raw) ash (mass %, dry) combustibles (mass %, dry) volatiles (mass %, daf) ultimate analysis (mass %, daf) C H O S N Cl lower heating value (MJ/kg daf) ash analysis (g/kg dry ash) K Na Al Si Fe Ca Mg P Ba trace elements (mg/kg dry ash) Hg Cd Pb Cr Cu Mn Co Ni As Sb V Ti Zn a

wood waste

8.7 0.5 99.5 81.5

38.0 0.8 99.2 82.1

50.3 6.1 43.5 0.010 0.09 0.010 18.6

49.8 6.1 43.8 0.014 0.22 0.030 18.0

87 8.1 15 75 16 184 33 13 2.6

69 27 12 77 51 129 23 6.0 1.5

3 21 112 106 413 25700 9 38 20 15 17 15 2800

3 17 660 945 1145 10750 22 125 392 20 43 20 6470

The ash analyses were made after ashing at 550 °C.

was less than two percent. No limestone was added for sulfur capture (except that of the fuel). The bed material was silica sand. The fuel was fed on top of the bottom bed. Just above that location, secondary air is added. Combustion takes place mostly in the bottom bed and, to some extent, in the riser. A minor part of the combustion occurs in the cyclone. The measurements were carried out downstream of the cyclone. The Fuel. The average composition of the fuels used during the tests is shown in Table 2. Minor variations in the concentrations may occur from case to case and from laboratory to laboratory engaged for the analyses. Some numbers can be given as examples: root-mean-square variations on S and Cl were 0.001% (daf) and on potassium, 5 g/kg dry ash. Actual demolition wood is not a well-defined fuel. To avoid uncontrolled variations in fuel composition, the well-defined fuels of Table 2 were selected to be combined with additives in order to simulate demolition wastes. The base fuel was high-quality waste wood (WW) from the industry, supported by wood pellets (WP) to satisfy the heat balance of the furnace. Both fuels are low-ash wood fuels, quite similar to each other, except for the high and somewhat variable moisture content in waste wood in contrast to the low and constant moisture content of wood pellets. Minor differences in the concentrations of trace elements between the two fuels are not of importance in the present context. For this reason, the pellets and the wood wastes can be regarded as one fuel. Combinations of these wood fuels served the purpose of maintaining the desired combustion conditions (particularly the bed temperature). To these fuels, alone or in combination, zinc oxide (ZnO) with a size of 20-25 µm was added in two test cases to represent pigment of white color that was previously used in buildings. The amount of zinc was such that it corresponds to a content of zinc in the ash of 125 g/kg dry ash (test case WP56 + ZnO) and 105 g/kg dry ash (test case WP51 + ZnO + HCl). In one test case, chlorine was introduced directly into the furnace

Ash Deposition in Combustion of Demolition Wood

Energy & Fuels, Vol. 20, No. 3, 2006 1003

Table 3. Molar Ratios in the Test Cases (Numbers Related to the Test Runs are Expressed in Mass % of Wood Pellets (WP), the Remainder Being Waste Wood) molar ratio runs

Cl/Zn

Cl/(K + Na)

S/Zn

2S/(K + Na)

(Al + Si)/(Na + K)

Ca/S

WP100 WP38 WP56 + ZnO WP51 + ZnO + HCl

22 4.9 0.9 4.0

0.2 0.1 0.3 1.9

34 6.1 0.7 0.7

1.1 0.3 0.4 0.7

1.4 0.9 1.3 1.6

3 10 7 4

Table 4. Cutoff Diameter (D50%) and Aerodynamic Diameter (Dpi) of the Particles Collecteda stage ELPI D50% (µm) ELPI Dpi (µm) DLPI D50% (µm) DLPI Dpi (µm)

1

2

3

4

5

6

7

8

9

10

11

12

13

0.029

0.059

0.103

0.166

0.255

0.393

0.637

0.99

1.61

2.46

3.97

6.69

10.15

0.041

0.078

0.131

0.206

0.317

0.500

0.794

1.26

1.99

3.13

5.15

8.24

-

0.029

0.059

0.104

0.167

0.256

0.395

0.641

1.00

1.62

2.47

4.00

6.73

10.21

0.041

0.078

0.132

0.207

0.318

0.503

0.801

1.27

2.00

3.14

5.19

8.29

-

aD 50% is the diameter whose collection efficiency is 50% in the stage concerned. Dpi is the geometrical average of the limits of each stage, Dpi,stage x ) (D50%,stage x × D50%,stage x + 1)1/2.

Figure 2. Particle sampling system with the two cyclones (CY1 and CY2) and the impactor (DLPI).

Figure 1. Deposition probe with a ring, whose surface deposits were analyzed.

through a spray nozzle, located in the fuel inlet, in the form of HCl that was instantaneously evaporated, well mixed with the fuel. The amount of chlorine injected corresponds to an increase in the chlorine content of the fuel from 0.02% (daf) to 0.156% (daf). The corresponding molar ratios are seen in Table 3. Test Strategy. Table 3 shows molar ratios of the fuels and additives in the four tests run. The table clearly shows that zinc is not important compared to chlorine, unless it is added in the form of contaminated fuel. Also, the small amount of sulfur in wood dominates over chlorine as far as interaction with potassium and sodium is concerned, if additional chlorine is not supplied. In wood, most alkali components are free to evaporate. They are not bound in minerals, predominantly aluminum and silicon, as in coal. However, as seen in Table 3, aluminum and silicon are available in sufficient quantities for possible reaction with sodium and potassium. Furthermore, the calcium in the fuel is sufficient to bind the sulfur if it is given the opportunity. Deposition Probe. The deposition probe was inserted downstream of the cyclone and upstream of the convection tube bundles into the center of the gas path at a flue gas temperature of 825 °C. It consists of an air-cooled tube, of 38-mm diameter, provided with a removable ring to collect deposits, as illustrated in Figure 1. The surface temperature of the ring was maintained at 500 °C. After exposure to the flue gases for 8 to 12 h, the deposit on the ring was removed for analysis. Hence, only initial deposits, rather evenly distributed around the ring, were concerned in this investigation.

Particle Measurements. The particles suspended in the gas flow were isokinetically sampled by a heated probe (kept at 120 °C) downstream of the heat exchanger surfaces of the convection path just upstream of the economizer at a temperature of 270 °C. The sampled gas-particle suspension was led to a low-pressure impactor (DLPI, Dekati low-pressure impactor) to measure its mass sizedistribution. The number size-distribution was continuously recorded by an electrical low-pressure impactor (ELPI). In these devices, particles are collected on foils, made of polycarbonate and covered with impactor foil grease (Apiezon L), in 13 steps according to Table 4. This equipment was also kept at 120 °C. To reduce the particle loading of the impactors, two cyclones were coupled upstream in series to remove the coarsest fractions, Figure 2. The cutoff and the aerodynamic diameters of the cyclones were D50% ) 10.2 and Dpi ) 17.5 µm (Cyclone CY1), and D50% ) 2.8 and Dpi ) 5.3 µm (Cyclone CY2), respectively. For CY1, the last particle size was assumed to be 30 µm. Analyses. A special effort was made to analyze the very small samples collected on the various impactor stages, 0.05-3 mg. The particles were collected on the foils in the impactor. After sampling, the foils with the collected particles were weighed. The samples were leached in water for 16 h, and then the contents of chlorine and sulfur were determined by ion chromatography. The water was evaporated, and the remaining sample was treated by hydrogen peroxide, fluorohydric acid, and nitric acid for 18 h at 85 °C. Then, the concentration of various species was determined either with ICP-OES (inductive coupled plasma-optical emission spectrometry, Perkin-Elmer/Optima 3000DV) or ICP-MS (inductive coupled plasma-mass spectrometry, Thermo Finningen/Element 2). This method can detect very small quantities of the species concerned, except for silicon (Si), whose lowest detectable quantity is 90 µg. Despite this high sensitivity, certain species could not be detected because of the very small samples. It should also be pointed out that the chemical analyses were much more sensitive than measurements of the mass concentrations of small particles: the limiting factor was the accuracy of weighing small quantities of captured particles together with the impactor foils. For this reason, the accuracy of mass concentrations below 1 mg/nm3 is not very high.

1004 Energy & Fuels, Vol. 20, No. 3, 2006

Åmand et al.

Figure 3. Deposit formation and composition on the test tube. The circle diagrams illustrate the relative concentration of compounds containing alkali, sulfur, chlorine, and phosphorus.

The relative concentration of selected critical compounds (KCl, K2SO4, K2CO3, CaSO4, CaCl2, NaCl, Na2CO3, Na2SO4, Ca3(PO4)2, ZnO, ZnCl2, and ZnSO4) were determined using a TOF-SIMS (timeof-flight second ion mass spectrometer, manufactured by Cameca/ IONTOF) instrument, equipped with three ion guns for analytical and sputtering purposes and a reflection TOF analyzer. This instrument carries out both surface and depth analyses on amorphous and nonamorphous materials. The instrument was calibrated against standard solutions in order to determine their concentration relative to KCl. A number of other compounds (K3PO4, FePO4, AlPO4) were also looked for, but no characteristic peaks from these compounds were observed on the mass spectrometer of the TOFSIMS. Generally, the analyses of alkali compounds and ZnCl2 were made with high accuracy. However, some compounds, such as ZnO and ZnSO4, produce a very weak signal, and the precision in their determination is lower, an estimate is (5 mass %. The data on these compounds, presented below, are the result of repeated analyses, which confirmed that the results were sufficiently accurate for the present purpose. The TOF-SIMS analyses were an average of four points, taken on the front side of the tube; furthermore, measurements were made on collected particles.

Results and Discussion Deposits on Tubes. The deposits collected on the test tube are shown in Figure 3. As seen from the bar diagram, the total amount of zinc is small in the deposits. Most of the deposits are related to potassium, chloride in the case of chlorine addition, and sulfate otherwise. During addition of zinc, the deposit was almost twice as large as for wood only, but still small. In this case, an important additional compound was ZnO, but an enhanced deposition of potassium sulfate was also observed. There was only a small fraction of ZnCl2. Table 3 reveals that the operation is in a safe regime as far as chlorine is concerned, because the molar ratio of 2S/Cl is about 6, and such a condition implies low corrosion according to experiences from waste combustion.8 Alternatively, the same could be stated if the molar (8) Krause, H. H. J. Mater. Energy Syst. 1986, 7 (4), 322-332.

ratio is (2S/available alkali) >4 according to Robinson et al.9 The alkali given by Table 3 is mostly “available alkali” because little alkali is bound in the fuel ashes of wood, as shown by leaching analysis (Yrjas et al.10). This criterion is less than 4, but the limit is not sharp, and the same conclusion as that of Krause can probably be drawn. This picture changes when chlorine becomes available: now KCl dominates and there are also other alkali compounds and some more ZnCl2. The situation can be suspected to be unfavorable with respect to corrosion. In all cases except that of chlorine addition, the deposits were light and easy to remove. The criteria of judgment are rough, and other explanations are possible in special cases. For instance, Table 3 shows that potassium can react with sulfur, but some potassium is left for chlorine to form KCl. If no chlorine is added to the furnace, there is a deficit of chlorine; even if chlorine reacts with potassium, the partial pressure of KCl will be below the saturation pressure at the temperature of the deposit tube and nothing condenses there. The tube is covered by potassium sulfate, which should be deposited as a powder at the temperature prevailing. However, as the gas is cooled further in the convection path, nucleation of KCl takes place before the gas reaches the sampling position of the impactor probe: submicrometer KCl particles are recorded, as will be shown below. When chlorine is added, on the other hand, the free excess potassium reacts to form KCl, its partial pressure becoming high and exceeding the saturation pressure of KCl: condensation can take place on the surface of the deposition probe. If condensation takes place at a higher temperature than that of the tube surface, nucleation occurs even upstream of the tube and diffusion of submicrometer KCl particles to the surface is then an additional mechanism of deposition. In the case of chlorine addition, chlorine takes care of the potassium, forming KCl. Then, judging (9) Robinson, A. L.; Junker, H.; Baxter, L. L. Energy Fuels 2002 16, 343-355. (10) Yrjas, P.; Skrifvars, B.-J.; Hupa, M.; Roppo, J.; Nylund, M.; Vainikka, P. In Proceedings of the 18th International Conference on Fluidized Bed Combustion; American Society of Mechanical Engineers: New York, 2005, Paper FBC2005-78097.

Ash Deposition in Combustion of Demolition Wood

Figure 4. Mass size-distribution of particles in the convection path. (Mass of particles m at size Dp.)

Figure 5. Number size distributions of particles in the convection path. (Number of particles N at size Dp.)

from the deposit on the tube shown in Figure 3, the sulfur is free to form CaSO4, which is now found on the tube together with KCl, in contrast to the cases without Cl addition when the deposit consists of K2SO4. Fly Ash. The measured fly ash particles may reveal something about the formation of deposits. The mass size-distributions of fly ash particles in the convection path, Figure 4, show the bimodal shape, characteristic of homogeneously condensed and eventually agglomerated submicrometer particles (Sarofim and Helble2; Valmari et al.11) in all cases, but particularly during chlorine addition. The corresponding number-size distributions in Figure 5 have maxima for very small particles, most likely caused by homogeneous nucleation. The size distributions are quite similar in all cases, except during addition of chlorine when it can be assumed that a very large amount of particles is formed. It can be speculated that the large number of particles leads to agglomeration, which reduces the number of very small particles and increases the number of larger particles even up to 1 µm. A view of the composition of the collected particles is given in Figure 6a-d. The diagrams comprise the major species and do not include trace elements. Therefore, the sum does not reach 100%. It is clear that, in all cases, but particularly during the addition of chlorine, KCl dominates the small particles, having a size below half a micrometer in all cases. This confirms that the submicrometer peak in the size distribution of Figure 4 is predominantly formed by alkali particles. There is less chlorine in the larger particle sizes; only during addition of chlorine to the combustor is some chlorine found in that size range. The (11) Valmari, T.; Lind, T. M.; Kauppinen, E. I.; Sfiris, G.; Nilsson, K.; Maenhaut, W. Energy Fuels 1999, 13, 379-389.

Energy & Fuels, Vol. 20, No. 3, 2006 1005

larger particles consist more of aluminum, silicon, and calcium, but potassium is always present to some extent, and combinations of potassium, silicon, and aluminum are most likely. Zinc is found in a narrow band in the range above 1 µm. An example of the distribution of the mass of some selected species in the fly ash is presented in Figure 7 for the case of wood (WP100). Potassium and chlorine follow each other in the range of small particles, but the potassium content increases for the larger particle sizes, whereas the chlorine content remains more or less the same also in the larger-particle size range. Sulfur is depleted in the small-particle range, perhaps because of deposition on the upstream tube bundles. Zinc is not important in the case of wood, but as far as there is any zinc, it is found in the large-particle range. The peak observed in the smallparticle range for all cases in Figure 4 can also be seen in Figure 8 representing the sum of K, Na, Zn, Cl, and S, species that obviously form most of this peak. In the HCl addition case, shown in the diagram, this sum almost entirely corresponds to the total sum of ash for small particles. In the large-particle range, on the other hand, these species are less dominant, and there is a gap between the total sum of ash and the sum of K, Na, Zn, Cl, and S. As was shown in Figure 6, here, aluminum, silicon, and calcium are more important. A survey of the distribution of zinc in the various size ranges of the fly ash particles is presented in Figure 9. The distribution agrees with the results of Lind et al.5 The concentration of zinc particles increases by an order of magnitude in the entire measured size range during zinc addition to the furnace compared with the case of wood only. The added zinc is transformed into chlorides and sulfates but also remains as an oxide. There is no peak indicating nucleation, such as in the case of KCl. The original size of the added zinc particles was 20-25 µm, so the finer particles seen in Figure 9 have undergone some transformations, including size reduction. The larger-particle range contains more traces from the injected ZnO, and in this range there were only minor fractions of ZnCl2, whereas the fraction of ZnCl2 is greater in the small-particle range in both the cases with and without chlorine injection. Chlorine injection increases the fraction of ZnCl2 in all particle sizes, but this is visible especially in the larger-particle range. Comments A few remarks can be made to further elucidate the results of the tests. Injection of Zinc. The width of the combustion chamber is about 1.5 m. The fuel is injected through a fuel chute, inclined with an angle of 45 degrees downward from the horizontal plane, assisted by air. It is estimated to hit the splash zone and the surface of the dense bed at about half the width of the furnace. The fuel may be partly immersed in the bed, partly remaining in the splash zone, but the fine zinc powder has certainly been carried away by the gas upward in the riser into the cyclone, following the gases out through the convection section to the final separators. In such a case it may not have been present in the possibly reducing zones of the bed, but it has been exposed to a predominantly oxidizing environment for about three seconds at a temperature of 800-850 °C before it meets the deposition probe. The Boiler. The three most common devices for combustion of wastes are circulating, noncirculating (“bubbling”) fluidized bed, and grate firing. The differences between the combustion devices with respect to zinc most likely consist in whether or not the zinc found a reducing region for reaction 1 to be effective, which would lead to gaseous zinc and finer particles,

1006 Energy & Fuels, Vol. 20, No. 3, 2006

Åmand et al.

Figure 6. Distribution of the main inorganic species on particle size: (a) WP100, (b) WP38, (c) WP56 + ZnO, and (d) WP51 + ZnO + HCl.

Figure 7. Mass distribution of K, Na, Zn, Cl, and S in the fly ash sampled in the convection path for wood pellets, WP100. (Mass of particles m at size Dp.)

even in the case of subsequent oxidation or chlorination. The dense bed of the noncirculating fluidized bed could promote such a behavior, but also the other devices, the fixed bed of a grate fired boiler and even the dense bottom bed of a CFB could be partly reducing (depending on the strategy of operation of the boilers). This might lead to finer zinc particles than those measured in the present case. However, Lind et al.5 obtained a size distribution of zinc in the fly ash similar to the present one during combustion of biofuels in a CFB, where much of the zinc was found in the bottom bed. The Potassium. Two sets of results have been shown: the deposits on a tube, whose temperature was 500 °C, located in a surrounding of 800 °C, and ash particles sampled at 270 °C.

Figure 8. Distribution of the sum K, Na, Zn, Cl, and S in the fly ash sampled in the convection path in the various cases investigated compared to the total mass in case WP51 + ZnO + HCl. (Mass of particles m at size Dp.)

Between these locations a certain deposition of ashes has taken place on the heat transfer surfaces between, and furthermore formation of ash particles has taken place from condensing vapors (formation of potassium particles should take place in the temperature range around 800 °C). Hence, it is not clear if the deposition of KCl on the deposition probe has been caused by condensation of supersaturated (in relation to the temperature of the thermal boundary layer) KCl on the surface of the tube, or if the deposition took place through diffusion of submicrometer particles, which were formed somewhere in the gas in the vicinity of the deposition tube.

Ash Deposition in Combustion of Demolition Wood

Energy & Fuels, Vol. 20, No. 3, 2006 1007

Figure 9. Distribution of zinc in the fly ash particles captured in the convection path. The circle diagrams describe the relative concentrations of zinc compounds. (Mass of particles m at size Dp.)

Conclusions In the temperature range investigated, presence of chlorine, leading to formation of KCl, is the main reason for substantial deposits on heat transfer surfaces during combustion of wood. The deposit of KCl forms by condensation on the tube or by deposition of submicrometer particles, if such particles have had time to form in the vicinity of the tube. At a lower temperature (270 °C), considerable amounts of submicrometer KCl particles were measured. During combustion of demolition wood contaminated with zinc, in addition to KCl, a small amount of ZnO and ZnCl2 is found in the deposits as well as in the fly ash particles. However, in the case of zinc, most particles are larger than 1 µm. The compounds mentioned, especially KCl, have some tendency to settle on heat transfer tubes, forming deposits that can contribute to corrosion. Only initial deposits on a single tube were studied in the present work. There are only small quantities of sulfur and chlorine in wood. Under normal conditions for wood combustion, sulfur dominates in the deposits, and this avoids chlorine components on tubes, although chlorine compounds in the gases condense at lower temperatures, forming submicrometer particles. Hence, wood alone or with contamination of zinc without chlorine is not a serious reason for deposits.

Zinc, which may accompany demolition wood, was added in the present tests as ZnO of a size of 20-25 µm, but zinc particles (mostly ZnO) were sampled from the flue gases in all size ranges above 1 µm, and some of the zinc had converted into ZnCl2, even at normal chlorine concentrations for wood, but without forming a submicrometer condensation mode similar to that of KCl. The amount of ZnCl2 found in the particle suspension was only slightly greater when chlorine was added, compared with the case of wood only. The CFB boiler was operated under normal conditions, and the overall combustion was well-behaved. Acknowledgment. This work was supported financially by the Swedish National Energy Administration. Operation of the boiler by personnel from Akademiska Hus, support in sampling and recollection of data from the technical staff of the Division of Energy Conversion, and the patient support in carrying out chemical analyses of the particle samples and deposits by Peter Sjo¨vall and Benny Lyve´n at the Swedish Testing and Research Institute are gratefully acknowledged. EF0503919