Formation of bed agglomeration in a fluidized multi-waste incinerator☆

Formation of bed agglomeration in a fluidized multi-waste incinerator☆

Fuel 82 (2003) 843–851 Formation of bed agglomeration in a fluidized multi-waste incineratorq Rong Yan*, David Tee Liang, Karin Lau...

292KB Sizes 0 Downloads 163 Views

Fuel 82 (2003) 843–851

Formation of bed agglomeration in a fluidized multi-waste incineratorq Rong Yan*, David Tee Liang, Karin Laursen, Ying Li, Leslie Tsen, Joo Hwa Tay Institute of Environmental Science and Engineering, Nanyang Technological University, Innovation Centre, Block 2, Unit 237, 18 Nanyang Drive, Singapore 637723 Received 12 November 2001; accepted 4 October 2002; available online 16 December 2002

Abstract A case study was carried out to investigate the bed agglomeration observed in a fluidized bed incinerator when burning blends of three wastes (carbon soot, biosludge and fuel oil). Several instrumental approaches were employed (i.e. XRF, SEM, XRD, and ICP-AES) to identify the bed materials (fresh sand and degrader sand) and clinkers formed in the full-scale incinerator tests. Several elements (V, Al, S, Na, Fe, Ni, P, and Cl), which normally are associated with the formation of low melting point compounds, were found in the waste blends at high content levels. The clinker bridges were identified to be associated with Al, Fe, V, K, Na, S, Ni, and Si elements. The effects of temperature and blending ratio were investigated in a muffle furnace. Carbon soot is believed to be more susceptible to the clinker formation than the other two fuels. Thermodynamic multi-phase multi-component equilibrium calculations predict that the main low melting point species could be Al2(SO4)3, Fe2(SO4)3, Na2SO4, NaCl, Na2SiO3 and V2O5. This information is useful to understand the chemistry of clinker formation. Also, it helps to develop methods for the control and possible elimination of the agglomeration problem for the design fuels. q 2002 Elsevier Science Ltd. All rights reserved. Keywords: Fluidized bed combustion; Agglomeration; Sintering; Clinker; Ash chemistry

1. Introduction Fluidized bed combustion (FBC) technology has emerged recently as the new, flexible, multi-fuel boiler for waste combustion and energy recovery from low-grade fuels. It offers such advantages as efficient combustion, ease of control, the ability to handle variable feeds, the ability to operate intermittently, and reasonable-to-low capital and operating costs. Additional advantages relative to hazardous wastes include fuel savings and lower emissions of nitrogen oxides (NOx) and metals [1 – 4]. Moreover, FBC is particularly useful due to the good mixing and relatively low operation temperature preventing many of the ashrelated problems, which might occur frequently in other types of boiler. However, ash-related problems can still lead to operational failures with FBC. A major potential problem encountered in fluidized beds, though, is bed sintering or agglomeration, which in the worst case may result in total defluidization often leading to unscheduled downtime. * Corresponding author. Tel.: þ 65-679-43244; fax: þ 65-679-21291. E-mail address: [email protected] (R. Yan). q Published first on the web via –

The word sintering encompasses many different processes, but generally can be said that it implies a process in which a porous substance through heating becomes more solid [5]. The main force for the particle coalescence or enlargement is surface energy minimization. Agglomeration, as sintering, also implies an increase in particle size but it is more a physical than a chemical process. Attention was paid on agglomeration of bed materials in fluidized bed combustion/gasification in the mid-seventies [6]. Since then, some fundamental research on agglomeration and defluidization has been carried out [7 –10], they found that agglomeration is caused mostly by sintering of coal ashes. For the agglomeration occurring in FBC of biomass, the ash chemistry is the most important. The potassium content is, apart from of course the temperature, the foremost sintering promoter [11,12]. It is generally believed that the elements influencing sintering and agglomeration the most are potassium, sodium, calcium, magnesium, silicon, sulfur and chlorine [13,14]. Vanadium is also believed to play an important part in bed agglomeration [15]. Nevertheless, recently Anthony et al. [4] reported that V is unlikely to be responsible for agglomeration because it was combined in the form of

0016-2361/03/$ - see front matter q 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 0 1 6 - 2 3 6 1 ( 0 2 ) 0 0 3 5 1 - 4


R. Yan et al. / Fuel 82 (2003) 843–851

high melting point Ca and Mg vanadates and not present as low melting point V2O4 or V2O5. The risk for sintering is also linked to the oxygen concentration [5]. Local oxygen surpluses can result in hot spots or zones that will induce sintering. The quality of bed mixing can therefore indirectly impact the sintering tendency as it will cancel out local temperature gradients. The bed material can, hence, indirectly influence the sintering behavior even if it is chemically inert as it acts as a temperature homogenizer. Two possible mechanisms for particle agglomeration in FBC are supposed [5]. In the first one, partial melting or reactive liquid sintering, ash with high sodium, chlorine and sulfur contents may form low-melting point eutectics, which partially melt at low bed temperatures, i.e. 500– 700 8C. The presence of the liquid phase makes the ash ‘sticky’ and would facilitate the transfer and adhesion of ash to the bed particles. The second mechanism, viscous flow sintering, involves melting of ash at temperatures greater than 1000 8C, producing a highly viscous liquid phase, which controls the sintering process. The high viscosity of the molten ash may keep the liquid in the glassy phase when the ash is deposited onto the bed particle surface whose temperature is normally lower than that of the char particle. A lot of researches have been carried out on the prediction of FBC failure due to ash-related problems. The ASTM/DIM standard ash fusion test is often employed

for predicting bed agglomeration [16]; another predictive method is the ash pellet compression strength measurement [17]. But both methods neglect the impact of combustion environment. Many FB rig and pilot plant tests are carried out both under reducing and oxidizing conditions [18,19]. Although test rig experiments are time consuming and results can seldom be transposed directly to full-scale equipment, these experiments are still cheaper than fullscale experiments and proved to be very valuable for understanding of sintering phenomena. Most recently, thermodynamic multi-phase multi-component equilibrium (TPCE) calculations are found to be a useful prediction tool for the formation of agglomerates in FB gasification of biomass fuel thereby enhancing the understanding of the chemistry involved [17,20]. This paper describes a case study on the bed materials agglomeration observed in a FBC system during its commissioning trials. One local company installed the FBC system to offer service for their clients in terms of waste collection and disposal by incineration. There are three main design wastes: carbon soot from a petrochemical plant; biological waste sludge and heavy fuel oil. It is expected to achieve the full capacity of 75 ton/day of the design wastes at the end of the commissioning period. Fig. 1 shows the multi-waste combustion system employed in this study. A detailed description about this system can be found in Section 2. Bed defluidization is particularly

Fig. 1. Multi waste combustion system.

R. Yan et al. / Fuel 82 (2003) 843–851

critical in the combustion of industrial wastes where the feed frequently contains high levels of the elements that are known to cause agglomeration. In order to understand the formation of bed agglomerates in the FBC reactor, a full investigation of the three waste feeds is conducted by several approaches, including scanning electron microscope (SEM), X-ray diffraction (XRD), X-ray fluorescence (XRF), and inductively coupled plasma atomic emission spectrometry (ICP-AES). A bench test is carried out in a muffle furnace to investigate the effect of temperature and feed blending ratio on the formation of agglomerates. Two types of sand are tested in the fullscale FBC reactor and the clinkers formed in both cases are also provided for analysis by SEM/XRF/XRD. For better understanding the ash chemistry involved, TPCE calculation is used to identify which phases could form and which composition the phases would have that lead to possible agglomerates, over a temperature range (400 – 2000 K) and at atmospheric pressure. This work is useful to understand the chemistry of clinker formation in the studied FBC system. Also, it helps in developing methods for the control and possible elimination of the agglomeration problem for the design fuels as well for any future waste fuel candidates.

2. Experimental section 2.1. Waste combustion system The waste combustion system (Fig. 1) consists of a fluidized bed reactor, a ventilation and pollution abatement system and multi-waste feeding system. The feeding system contains three inlets for granular solid, sludge and fuel oil. The multi-waste combustor, a FB reactor, is heated up to 500 – 600 8C and the waste is introduced continuously into the reactor and immediately mixed into the hot sand mass (the sand particle size is around 300 mm). A good fluidization can be achieved in the studied FB reactor if operation follows the specification. The operational temperature (500 – 600 8C) is designed quite low in preventing ash agglomeration at high temperatures; it permits the FB reactor to combust a wide variety of wastes. The water contained in the waste is evaporated, and the organic components are gasified and degraded while the inorganic compounds break down into very fine inert particles, which are blown out of the fluidized bed reactor. The big particles are retained in the FB bed mix with the bed sand and after a certain period the bed is turned over. In the post-combustion chamber, the gases are heated up above 850 8C or higher under strongly oxidizing atmosphere, in order to guarantee a complete oxidation of all organic compounds. Acidic gases (such as SOx, NOx and HCl) and metal vapors are removed by sprayed lime and activated carbon powders, whilst solid particles are captured by the ceramic filter.


2.2. The fuels studied The three feed wastes, carbon soot, biosludge and fuel oil, were provided for analysis by Bomb Calorimeter and CHNS Elemental Analyzer after drying, as well by XRF and ICP-AES after low-temperature ashing. The composition of the fuels as received is given in Table 1. Thirty-five tons of waste is being processed per day as a mixed waste with 62% carbon soot, 20% biosludge and 18% fuel oil. The bed turnover rate is estimated to be about 10 –12 days/bed as the original sand is around 12 –15 ton. High ash contents and a large quantity of heavy metals are found in both carbon soot and biosludge, whereas only a few of them are detected in fuel oil. Extremely high vanadium (23.2%) is found in carbon soot, while other major elements are Ca, Fe, Na, Ni and S with concentration more than 1%. In biosludge ash, Al is the most significant element (31.2%) and the concentrations of Cl, Fe, Na, P and S are also more than 1%. 2.3. Muffle furnace test The three fuels were burned together with the bed material, sand, in a muffle furnace with the temperature ranging 500 –900 8C and under a static ambient atmosphere. The crucible containing sand and fuels was put into the furnace. The furnace was first heated at the preset ramp and the temperature was rapidly increased to the desired value (500 –900 8C), generally within 5 min. Then, the temperature was kept for 5 h to complete the combustion. Both the single fuel and blends of the three at different ratios were considered in order to find out the effects of fuel varieties on the clinker formation. The formed clinkers were picked out from the crucible and provided for identification using SEM, XRD and XRF. At the experimental condition, air was insufficient and the reaction atmosphere was more close to a reducing condition. 2.4. SEM/XRF/XRD/ICP-AES analysis With SEM, XRF, XRD and ICP-AES, a lot of analyses were carried out for different samples, including two types of fresh sand, degrader sand, clinkers formed from two fullscale tests in the FBC reactor with different sands, three fuels (dried and ashing samples), clinkers formed from muffle furnace tests, etc. The SEM uses electrons to form a 3-dimensional image of the outer surface of the test materials and indicates the elemental distribution in detail. It shows the bonding image of sand particles in the clinker and the chemical composition of the clinker bridge can also be obtained if combining SEM with energy dispersive X-ray spectroscopy (EDS) technique. XRF is a spectroscopic method that is widely used to measure the elemental composition of material; the solid samples can be tested rapidly without acidic digestion. The XRD unit is used to identify organic and inorganic


R. Yan et al. / Fuel 82 (2003) 843–851

crystalline chemical compounds in a sample, especially the possible crystalline transformation during clinker formation. ICP-AES is used to determine the concentrations of trace elements presented in samples after the sample digested by acidic solution. The results and discussions obtained from the above-mentioned analyses are provided in the subsequent text.

the stoichiometric reaction between fuel and oxygen: Ca Hb Sg Nd Ow ·ðH2 OÞm ðcrÞ þ ða þ b=4 þ g þ d=2 2 w=2Þ O2 ðgÞ ! aCO2 ðgÞ þ ðb=2 þ mÞH2 OðgÞ þ gSO2 ðgÞ þ dNOðgÞ ð1Þ the stoichiometric air requirement (theoretical value) is given by:

2.5. TPCE calculation

Lmin ¼ ða þ b=4 þ g þ d=2 2 w=2Þ=YðO2 Þ

TPCE calculation is conducted to understand better the ash chemistry involved during combustion of the three fuels blend, particularly the information about the chemical combination of the concerned elements. The principle of TPCE is to minimize the total Gibbs free energy of the system. GEMINI code and COACH database [21] were employed in this study. To simulate the actual condition of fuel combustion in the fluidized bed combustor, the air excess number (l) is calculated assuming the following general fuel molar composition: CaHbSgNdOq·(H2O)m, and


where Y(O2) is the mole fraction of oxygen in air. The air excess number (l) used in the calculation is defined by:

l ¼ L=Lmin


where L is the real air supplied. l values for combustion under reducing and oxidizing atmospheres are set, respectively, to 0.6 and 1.4 in this computation work; the above supposition decides the combustion atmosphere condition. Three fuels are blended in the operational ratio and bed sand is also put into

Table 1 Composition of the studied fuels Fuels

Carbon soot (wt%)

Biosludge (wt%)

Fuel oil (wt%)

Total (kg/day)

Weight/daya Water content Water per day introduced Dry weight per day Ash content Ash produced per day (in dry basis) C H O N S (in ash) Al Br Ca Cl Co Cu F Fe K Mg Mn Na Ni P Rb S Se Si Ti V Zn Zr Sand–SiO2

21,700 kg 84.26% 18,284 kg 3416 kg 27.14% 927.1 kg 60.22% 5.395% 0.249% 6.374% 0.622% 0.717% 0.18% 1.03% 0.49% 184 ppm 227 ppm 0.66% 10.8% 0.514% 0.596% 338 ppm 4.02% 7.23% 0.155% 0.593% 2.96% 18.5 ppm 392 ppm 207 ppm 23.2% 634 ppm 0.318% 12,000 kg

7000 kg 88.33% 6183 kg 817 kg 57.68% 477.9 kg 40.8% 0.795% 0.0 0.227% 0.5% 31.2% 0.14% 0.507% 3.95% 0.213% 149 ppm 0 2.28% 0.390% 0.151% 0.152% 4.86% 808 ppm 11.4% 0.380% 6.6% ,1.1 ppm 0.185% 29.6 ppm 67.9 ppm 0.620% 0.096%

6300 kg 0.0 0.0 6300 kg 0.09% 5.67 kg 87.67% 10.82% 0.0 0.327% 1.09% 11 ppm – – 0 – 3.8 ppm 0 6.7 ppm ,12 ppm ,33 ppm – ,82 ppm 27 ppm ,9.4 ppm – 0.941% – 4.6 ppm 5.6 ppm 23 ppm ,3.2 ppm –



35 ton of waste is being processed per day as a mixed waste with 62% carbon soot, 20% biosludge and 18% fuel oil.

24,467 10,533 1410.7 8006.9 872.4 8.5 240.2 94.0 277.76 7.293 39.23 49.01 2.37 0.921 22.55 387.6 20.75 21.59 2.40 177.03 247.8 98.43 23.36 214.31 0.0632 2.879 0.766 792.7 7.26 11.65

R. Yan et al. / Fuel 82 (2003) 843–851


Table 2 Input molar number for TPCE calculation

l ¼ L=Lmin









1.4, 0.6 Si 200.1 Ti 0.016

9341.3, 4013.2 Al 10.29 Co 0.0402

991.9, 923.6 Ca 0.981 Cu 0.0145

2970.7, 1502.0 Mg 0.90 Ni 4.22

667.2 K 0.532 Rb 0.273

9.64 Na 7.697 Se 0.0008

1.187 Fe 6.921 V 15.54

1.38 Mn 0.0436 Zn 0.111

0.091 P 3.175 Zr 0.128

the system, although at low temperature it is to be regarded as inert material. Based on the composition of fuels as given in Table 1 and of air (78% N2, 21% O2 and 1% H2O), the molar number for each element input is calculated and listed in Table 2. The considered temperature ranges 400– 2000 K and pressure is 1 atm.

employed in the FBC reactor to understand the impact of different sands on bed agglomeration; however, clinkers were still formed in this test. XRF results for this new sand (Table 3) show that its main component is silica (. 98%) and it includes less Al, Fe and K than the previous sand. The Indonesia sand has the same particle size (around 300 mm) as the fresh sand.

3. Results and discussions

3.1.2. SEM –EDS analysis Back scattered electron (BSE) images (Fig. 2) of an epoxy embedded, cross-sectioned and polished part of the FBC bed agglomerate show that the bonding between the quartz particles is very strong. There is a clear reaction between quartz grains and the bridge that bonds the particles. The chemical composition of the bridge, bonding portion, shows a high content of elements (Table 4), which normally are associated with low melting point compounds (e.g. Al, Fe, V, K and Na). The high Si content may be an effect from the surrounding quartz grains. It is suggested that the formation of low melting point compounds might be the main route leading to the clinker formation in the studied FB reactor since quite low operational temperatures were

3.1. Analyses of fresh sands, degrader sand and clinkers 3.1.1. XRF analysis Results from XRF analysis of the clinker, degrader sand (sampled during a test run where no serious agglomeration was seen) and the fresh bed sand are given in Table 3. It shows that the main component is silica (. 92%). Other major elements presented in all samples include Al, Fe, K and Na. If compared with the fresh sand, the exhausted bed materials (including clinker and degrader sand) are enriched in Al, Fe, V and Ni elements. Between the clinker and the degrader sand, differences can be found related to the concentrations of S, P, Fe and Al. However, the chemical compositions of the two exhausted bed materials are similar in terms of the concentration of some elements (e.g. V, Na, K) which normally are associated with the formation of low melting point compound and clinker formation in fluidized bed combustion. Nevertheless, not only the content but also the chemical speciation of the element decides the physical behavior of bed materials such as melting. Although they were found at the same level of concentration in the clinker and the degrader sand, these elements (V, Na, K) may be present in different forms. Information about the chemical combination of these concerned elements will help to better understand the chemistry of the observed agglomeration. The TPCE calculation is thus conducted to obtain this information and the results will be presented in the subsequent text. In addition, a lot of other parameters (such as combustion condition, fluidization, velocity, etc.) also influence the bed status. Although no significant agglomerate was observed in the case of degrader sand, the concentration of the elements (V, Na, K) could possibly be the same level with those found in the clinker since the same waste feeds were used in the two cases. A new bed sand (Indonesia sand) was also

Table 3 XRF analyses of clinker, degrader sand and fresh bed sand (wt%) Sample


Degrader sand

Fresh bed sand

Indonesia sand

SiO2 Al2O3 Fe2O3 V2O5 K2 O Na2O BaO NiO CaO Cl TiO2 ZrO2 CuO SO3 MnO Cr2O3 ZnO P2O5 MgO SrO CoO Total

94.4 1.94 0.92 0.767 0.709 0.40 0.26 0.180 0.13 0.07 0.05 0.0375 0.014 ,0.029 0.015 ,0.011 0.0043 ,0.037 0.14 0.0037 0 100

91.8 3.0 1.71 0.697 0.591 0.50 0.25 0.198 0.14 0.069 0.042 0.111 0.015 0.23 0.025 ,0.010 0.033 0.59 0.14 0.0036 0 100

96.4 1.44 0.28 0 0.70 0.43 0.27 ,0.0051 0.12 0.10 0.045 0.116 0.013 ,0.028 0.0077 ,0.011 0.0063 ,0.039 0.15 0.0038 0 100

98.27 0.066 0.0035 – 0.031 0.75 0.18 0.0028 0.055 0.232 0.021 0.0556 0.0097 0.025 – 0.0042 – – 0.12 – – 100


R. Yan et al. / Fuel 82 (2003) 843–851

Fig. 2. SEM image of the embedded clinker.

used (500 – 600 8C). These compounds partially melt at low bed temperature; the presence of the liquid phase makes the ash ‘sticky’ and the adhesion of ash to bed particles occurs consequently. Although all clinker bridges carry low melting point materials (Table 4), the viscous flow sintering might still be a possible route for the formation of bed agglomeration since the local over-heat environment possibly existed in bed. Actually, the local industrial partner monitored the bed temperature at several specific zones during combustion and they confirmed the existence of local over-heat zones (800 – 900 8C) when the bed particles started to agglomerate.

lower this temperature significantly. It was previously supposed that the clinker probably formed at high temperature following the mechanisms of viscous flow sintering; crystalline transformation from quartz to cristobalite and tridymite might be an indicator of the ash agglomeration. However, heating of the degrader sand and the bed sand in a Muffle furnace for 2 h at 800, 1000 and 1200 8C respectively, did not transform the quartz to cristobalite in any of the samples. Nevertheless, slight agglomeration was observed in the two degrader sand samples heated at 1000 and 1200 8C; the bonding of the sample heated at 1200 8C was very strong. The formation of agglomeration from degrader sands at higher temperature is supposed to follow the viscous flow sintering. The transformation of mineral from quartz to cristobalite and tridymite was observed only in this clinker. The main minerals found in other clinkers formed from the tests using Indonesia sand in full scale and from muffle furnace tests are still quartz. Therefore, it is suggested that the formation of cristobalite and tridymite is not essential for FB bed agglomeration observed in this study.

3.1.3. XRD analysis The only mineral detected by XRD in samples of the fresh bed sand and the degrader sand is quartz (SiO2), whereas XRD analysis of the clinker shows that the quartz is transformed to the high-temperature form: cristobalite and tridymite. The transformation temperature of pure quartz to tridymite is around 800 8C and to cristobalite around 1400 8C. The presence of foreign cations (e.g. Na) may Table 4 FBC bed agglomerates by SEM–EDS Location

SiO2 Al2O3 Fe2O3 SO3 K2 O Na2O V2O5 Cr2O3 NiO Total

Sand grain

Bonding portion











97.84 0.67 0.36 0.38 0.02 0.39 0.26 0.05 0.03 100

97.66 0.77 0.40 0.31 0.09 0.46 0.34 0.00 0.03 100

97.35 0.91 0.37 0.17 0.15 0.52 0.32 0.03 0.19 100

99.33 0.13 0.11 0.27 0.00 0.26 0.07 0.01 0.00 100

98.05 0.62 0.31 0.28 0.07 0.41 0.25 0.02 0.06 100

83.57 6.96 2.98 0.17 2.77 1.24 2.13 0.02 0.16 100

83.19 7.05 3.15 0.24 2.71 1.29 2.33 0.00 0.10 100

83.51 7.17 2.73 0.33 2.79 1.36 2.00 0.04 0.07 100

84.92 6.50 2.64 0.32 2.39 1.10 1.88 0.07 0.16 100

83.80 6.92 2.88 0.27 2.67 1.25 2.09 0.03 0.12 100

R. Yan et al. / Fuel 82 (2003) 843–851


3.2. Muffle furnace tests The results of muffle furnace tests are given in Table 5. A lot of agglomerates were found when burning carbon soot together with sand at temperatures above 600 8C. Although for carbon soot no test was conducted at 800 and 900 8C, it is supposed that clinkers are to be formed significantly at these temperatures since higher temperature generally increases the risk of bed sintering. On the contrary, almost no clinker was observed when burning the other two fuels separately with sand, as observed in the cases of biosludge at 600 and 700 8C and fuel oil at 600 and 800 8C. It is believed that carbon soot is more susceptible to the clinker formation than other two waste fuels. Furthermore, when burning the mixed wastes only a few clinkers were observed in the temperature range 600 – 800 8C, whilst quite a lot formed at 900 8C. High temperatures always cause more risks in the formation of melting compounds; that is the reason why the FB reactor is designed to operate at quite low temperatures (500 – 600 8C). The ratios of sand/fuel were found insignificant since the sand is inert and also non-melting occurs at the studied temperatures. As seen in Table 5, two ratios (a and d) were employed for the cases of sand plus single fuel. When mixing fuels were involved, one specific fuel composition was employed (c) at 600 8C whilst the designed operation ratio (b) was mostly used at variable temperatures. The results indicate that the influence of ratios of sand/fuel is negligible in the muffle furnace case. Nevertheless, in a full-scale reactor the sand can influence indirectly the sintering behavior even if it is chemically inert as it acts as a temperature homogenizer. These clinkers were provided for SEM/XRD analysis. An example of the SEM images of the clinkers formed from the muffle furnace test is given in Fig. 3. Table 6 shows the chemical composition of the bridges in the five clinkers formed from muffle furnace tests. Different to the observation from previous SEM results (Fig. 2), the composition of the bridge was not homogenous; significant

Fig. 3. Morphologies of the clinker formed from sintering the three fuels blend at 900 8C.

differences were found between dark and bright areas (Fig. 3). This has probably resulted from the poor mixing of fuels with air in the muffle furnace. Average compositions of the clinker bridges (Table 6) indicate that extremely high contents of vanadium (V) found in the clinkers formed from burning carbon soot or mixed fuels, whereas calcium was rich in the clinker formed from sintering biosludge. Comparing with those samples sintered at lower temperatures, concentration of Na and S in the clinker formed at 900 8C decreased significantly. In addition, silica, which was always found in the clinker bridges from sample sintering at different conditions, was also believed to contribute to the bed agglomeration, as found by Steenari in 1998 [22]. Only quartz was found by XRD in these clinkers formed from muffle furnace tests. No transformation to cristobalite and tridymite occurs as observed in the previous clinker. 3.3. TPCE calculation

Table 5 Results of muffle furnace test Temperature (8C)

Carbon soot þ sand

Biosludge þ sand

Fuel oil þ sand

Mixture of fuels þ sand

500 600 700 800 900

£a UUUa UUUUd – –

– £a Ud – –

– £a – £d –

£b Uc; Ub Ub Ub UUUb

£ : no clinker found; U: a few clinker found; UU: a lot of clinker found; –: not tested. a , 80 g sand þ , 8 g single fuel, no mixing. b Three fuels mixed well in the operational ratio, , 12 g mixed fuels þ , 18 g sand. c 67.7 g sand þ 5.3 g carbon þ 4.37 g biosludge þ 4.4 g fuel oil, no mixing. d , 18 g sand þ , 12 g single fuel, no mixing.

At the real combustion conditions, i.e. the blending ratio of three fuels is carbon soot (62%), biosludge (20%) and fuel oil (18%), temperature ranges 500 –600 8C and at atmospheric pressure, the main species of those elements relative to ash formation and their temperature range are listed in Table 7. These species are predominantly presented in the multi-phases and multi-components combustion system in the view of thermodynamic equilibrium. Among the main species listed in Table 7, some are those compounds with low melting point, which may partially melt at low bed temperature (500 – 600 8C) thus causing bed material sintering. The following are their formula and melting points (in bracket), Al2(SO4)3 (770 8C), CaCO3, aragonite (825 8C), Fe2(SO4)3 (480 8C), CoSO4 (880 8C), CuSO4 (600 8C), MnSO4 (700 8C), MnCl2 (650 8C), MnO2


R. Yan et al. / Fuel 82 (2003) 843–851

Table 6 Chemical composition of clinker bridges Clinker source

Main composition (average)

Minor composition (average)

Difference between dark and bright areas

Carbon soot sintered at 600 8C Carbon soot sintered at 700 8C Biosludge sintered at 700 8C Mixed fuels at 600, 700 and 800 8C Mixed fuels at 900 8C

V and Si V, Si, Fe, Ni Ca, S, Si V, Si, Na and S Si, V, Fe, Ni

Na, S, Ca, Fe, Ni, K Al, Na, Mg, K Ca Al, Fe Ca, Fe, Ni, Al, K Na, Al, K, Ca

Bright area is rich in V Relatively homogenous – Dark area contains more Na and S than the bright area Bright area contained more V, Fe and Ni, while the dark was rich in Si

Table 7 Main species of mineral elements during combustion, by TPCE calculation (s: solid, l: liquid, g: gas, cr,l: condensed) Element

Oxidizing condition ðl ¼ 1:4Þ

Reducing condition ðl ¼ 0:6Þ


Al2O3(s) (400–2000 K), KAl(SO4)2(s) and Al2(SO4)3(s) (,900 K) CaSO4(s) (400–1400 K) and CaO(s) (400-2000 K) CoSO4(s) (,900 K), CoFe2O4(s) (800-1100 K) CuSO4(s) (,900 K), CuSO4.H2O(s) (,450 K)

Al2O3(s) (400–2000 K), KAl(SO4)2(s) (,900 K)

Ca Co Cu Fe K Mg Mn Na Ni P Rb Se Si Ti V Zn Zr

Fe2O3(s) (400–2000 K) and Fe2(SO4)3(s) (,800 K) KAl(SO4)2(s) (,900 K), K2SO4(s) and K2Cl2(g) (,1000 K) Mg(OH)2(g) (,1300 K) MnSO4(s) (500–1100 K), MnO2(s) (,500 K) Na2SO4(s) (,1500 K) and Na2SiO3(s) (,1400 K) NiSO4(s) (,900 K), NiFe2O4(s) (800-1300 K) H3PO4 (cr,l) (,500 K), Na2P2O6 (s) (500-600 K), Mg3P2O8(cr,l) (600–1200 K) Rb2SO4(cr,l) (,1000 K), RbCl(g) ( . 800 K) RbBr(g) ( . 900 K) SeO2(g) (,800 K), CoSeO3(cr,l) (,450 K) SiO2(s) (400–2000 K) TiO2(s) (400–2000 K) V2O5(cr,l) (,900 K), CaV2O6(s) (1000 K), CaV2O7(s) (1100–1200 K) ZnSO4(s) (,800 K), ZnAl2O4(s) (800–1400 K) ZrSiO4(s) (400–1700 K)

(. 230 8C), H 3PO 4 (42 8C), Na 2SO 4 (884 8C), NaCl (800 8C), V2O5 (800 8C) and Na2SiO3 (alkali metal silicate, 635 – 815 8C). In addition, K2SO4 (arcanite) might be transited to form K2S2O7 at 588 8C and the latter has melting point low to 300 8C. NiSO4 is possibly decomposed at 840 8C to Ni2O3 with low melting point (400 8C). Several species, i.e. Al2(SO4)3, Fe2(SO4)3, Na2SO4, NaCl, Na2SiO3 and V2O5, could be the main sintering promoters since their concentrations are relatively higher than other low melting point species in the combustion system. These species, the possible main sintering promoters, are presented mostly under oxidizing conditions, except for NaCl(s) that is found predominantly under reducing conditions. Therefore, the reducing atmosphere conditions may possibly restrain the formation of major low melting point species in the studied incinerator, thus reducing the possibility of bed material sintering. Note that at the computation condition, V is mainly in the form of low melting point V2O5, while at higher temperature (. 700 8C)

CaCO3[a] (s) and CaCO3[c] (s) (400–2000 K) Co3S4(s) (,1000 K), CoS0.89(s)(1000 K) Cu5FeS4(s) (450– 850 K), CuFeS2(s) (,450 K), Cu2S(cr,l) (800– 1400 K) FeO(s) (,1700 K) KAl(SO4)2(s) (,900 K) and KCl(s) (900–1500 K) Mg(OH)2(g) (,1300 K) MnO(s) ( . 600 K) and MnCl2(s) (,600 K) NaCl(s) (,1100 K) and NaFeO2(s) (800–1600 K) NiS0.84(s) (,750 K) Ca3P2O8(s) (,1000 K), K3PO4(s) (around 900 K) Rb2SO4(cr,l) (,500 K), RbBr(cr,l) and RbCl(cr,l) (,700 K), RbCl(g) and RbBr(g) ( . 550 K) H2Se(g) (,1400 K), PbSe(cr,l) (,700 K) SiO2(s) (400–2000 K) TiO2(s) (400–2000 K) V3O5(s) (,550 K), V2O3(cr,l) (500–1200 K) ZnS(cr,l) (,1000 K) ZrO2(cr,l) (,450 K), ZrSiO4(s) (450–1700 K)

and under oxidizing condition it presents as high melting point Ca vanadates, different to the Published Work [4] as mentioned previously.

4. Conclusions Based on the results from experimental measurement and TPCE calculation, the possible factors causing the formation of clinkers are drawn out: 4.1. High contents of V, Al, S, Na, Fe, Ni, P, and Cl are found in the blends of the three fuels Some of these elements are believed to influence sintering and agglomeration the most, i.e. V, S, Na and Cl, which normally are associated with formation of low melting point compound and clinker formation in fluidized bed combustion. On comparing the concentrations of these

R. Yan et al. / Fuel 82 (2003) 843–851

elements between carbon soot and biosludge, the extremely high content of V (23.2%), Fe (10.8%) and Ni (7.23%) in carbon soot (Table 1) accounts for the significant clinker formation from sintering carbon soot and sand in the muffle furnace tests. It might also be the main cause of the bed agglomeration occurring in the FB reactor. The identification of clinker bridges using SEM – EDS (Tables 4 and 6) also indicates that Al, Fe, V, K, Na, S, Ni, and Si are enriched in the bridges. They are normally associated with the formation of low melting point compounds. Because the operation bed temperature is quite low (500 – 600 8C), the formation of clinker normally follows the first mechanism, partial melting or reactive liquid sintering. 4.2. The main low melting point species formed during incineration of the three fuels blend are possibly Al2(SO4)3, Fe2(SO4)3, Na2SO4, NaCl, Na2SiO3 and V2O5 To simulate the real combustion condition, the main thermodynamically stable species are predicted by using TPCE calculation. Among them, six species, Al2(SO4)3, Fe2(SO4)3, Na2SO4, NaCl, Na2SiO3 and V2O5, are found to be associated with the bed sintering the most since they are all low melting point compounds and they have relatively high contents in the combustion system. Five of the six species (except for NaCl) are predominantly presented under oxidizing condition. It suggests that reducing conditions might be favorable for restraining the bed sintering in the studied FBC system. 4.3. Local over-heat environment existed in bed Lots of clinkers were found at 900 8C from the muffle furnace tests with the mixed fuels and sand; thus we suppose that the local over-heat situation in bed is susceptible for bed agglomeration. In the real combustion system, the fuel oil might ignite first and burn out more rapidly than other two fuels (carbon soot and biosludge), thus inducing a relatively high temperature zone and the formation of liquid phase that makes the ash ‘sticky’ and followed by particle size increasing and bed sintering.

at low temperature (600 8C) when sintering carbon soot alone with sand in the muffle furnace. In conclusion, the main factors causing the clinker formation in the studied FBC reactor should be the waste fuels employed, especially the carbon soot which contains high content of V, S, Na and Cl elements normally associated with the formation of low melting point compounds. An optimum blending ratio of the three wastes will help in restraining the sintering. Moreover, suitable combustion conditions are crucial to get a better mixing of fuels and to avoid the local high temperature zones possibly occurring in the FBC reactor bed.

References [1] [2] [3] [4]

[5] [6]

[7] [8] [9] [10] [11] [12] [13]


[15] [16] [17] [18]

4.4. High risk of agglomeration might also occur in the area where mixing is not perfect Although only a few clinkers were found at 600 8C in muffle furnace tests when three fuels were well mixed with sands (Table 5), operational feeding condition cannot guarantee a uniform distribution of the three fuels. Particularly, current feeding rate of carbon soot is quite high (62%) and a lot of clinkers have already been observed




[21] [22]

Oppelt ET. JAPCA 1987;37(5):558–86. Stevenson EM. Environ Sci Technol 1991;25(11):1808–14. Linak WP, Wendt JOL. Prog Energy Combust Sci 1993;19:145–85. Anthony EJ, Jia L, Preto F, Iribarne JV. In: Preto FDS, editor. Proceedings of the 14th International Conference on Fluidized Bed Combustion, vol. 2. ASME; 1997. p. 839–46. Skrifvars BJ, Hupa M, Backman R, Hiltunen M. Fuel 1994;73(2): 171–6. Gluckman MJ, Yerushalmi J, Squires AM. Defluidization characteristics of sticky materials on agglomerating bed. In: Kearins DL, editor. Fluidization technology. ; 1976. p. 395 –422. Basu P. Can J Chem Engng 1982;60:791–5. Siegell JH. Powder Technol 1984;38(1):13–22. Compo P, Pfeffer R, Tardos GI. Powder Technol 1987;51:85– 101. Arastoopour A, Huang CS, Weil SA. Chem Engng Sci 1988;43: 3063– 75. Ergu¨denler A, Ghaly AE. Biomass Bioenergy 1993;4(2):135–47. Ghaly AE, Ergu¨denler A, Laufer E. Boimass Bioenergy 1993;5: 467–80. Carty R, Mason DM, Babu SP. Reaction kinetics and physical mechanisms of ash agglomeration. Final Report, Doe/MC/21313Doe/MC/2593; 1988. Moilanen A, Nieminen M, Sipila¨ K, Kurkela E. Ash behaviour in thermal fluidised-bed conversion processes of woody and herbaceous biomass. VTT Energy, Espoo Finland; 1996. Jones C. Power 1995;46–56. DIN 51 730, Bestimmung des Asche-Schmelzverhaltens, Berlin; 1984. Zevenhoven-Onderwater M, Backman R, Skrifvars BJ, Hupa M. Fuel 2001;80(10):1489– 502. ¨ hman M, Hupa M, Nordin A, Liliendahl T, Rao A. Natarajan E, O Biomass Bioenergy 1998;15:163–9. Liliedahl T, Kusar H, Rosen C, Sjostrom K, Second Olle Lindstrom Symposium on renewable energy, bioenergy, Stockholm, Sweden: Royal Institute of Technology; 1999. p. 88–91. Zevenhoven-Onderwater M, Backman R, Skrifvars BJ, Hupa M, Liliedahl T, Rosen C, Sjostrom K, Engval K, Halgren A. Fuel 2001; 80(10):1503 –12. Yan R, Gauthier D, Flamant G. Combust Flame 2001;125(1– 2): 942–54. Steenari BM. Chemical properties of FBC ashes. Chalmers Dissertations. Chalmers University of Technology; March 27, 1998.