Evaluation of slagging and fouling tendency during biomass co-firing with coal in a fluidized bed

Evaluation of slagging and fouling tendency during biomass co-firing with coal in a fluidized bed

b i o m a s s a n d b i o e n e r g y 3 9 ( 2 0 1 2 ) 1 9 2 e2 0 3 Available online at www.sciencedirect.com http://www.elsevier.com/locate/biombioe...

2MB Sizes 0 Downloads 151 Views

b i o m a s s a n d b i o e n e r g y 3 9 ( 2 0 1 2 ) 1 9 2 e2 0 3

Available online at www.sciencedirect.com

http://www.elsevier.com/locate/biombioe

Evaluation of slagging and fouling tendency during biomass co-firing with coal in a fluidized bed Paula Teixeira a,*, Helena Lopes a, Ibrahim Gulyurtlu a, Nuno Lapa b, Pedro Abelha a a b

LNEG/UEZ, Estrada do Pac¸o do Lumiar, 22, Ed. J, 1649-038 Lisboa, Portugal UNL-FCT-DCTB-UBiA, Quinta da Torre, 2829-516 Caparica, Portugal

article info

abstract

Article history:

Over the last decades, several indices based on ash chemistry and ash fusibility have been

Received 20 April 2010

used to predict the ash behaviour during coal combustion, namely, its tendency for slag-

Received in revised form

ging and fouling. However, due to the physicalechemical differences between coals and

2 September 2011

biomass, in this work only the applicability of an ash fusibility index (AFI) to the

Accepted 5 January 2012

combustion and co-combustion of three types of biomass (straw pellets, olive cake and

Available online 3 February 2012

wood pellets) with coals was evaluated. The AFI values were compared with the behaviour of ash during combustion in a pilot fluidized bed and a close agreement was observed

Keywords:

between them. For a better understanding of the mechanisms associated with bed ash

Ash

sintering, they were evaluated by SEM/EDS and the elements present on the melted ash

Slagging

were identified. Evidences of different sintering mechanisms were found out for the fruit

Fouling

biomass and herbaceous biomass tested, depending on the relative proportions of prob-

Co-firing

lematic elements. The particles deposited on a fouling probe inserted in the FBC were

Biomass

analyzed by XRD and the differences between the compounds identified allowed

Fluidized bed

concluding that the studied biomasses present different tendencies for fouling. Identification of KCl and K2SO4 in the deposits confirmed the higher tendency for fouling of fruit biomass tested rather than wood pellets. ª 2012 Elsevier Ltd. All rights reserved.

1.

Introduction

Biomass combustion is gaining increased importance and is the subject of extensive R&D with the aim of reducing CO2 emissions and diversification of fuel sources. There are particular aspects of biomass combustion that need to be better understood. The on-going research areas include greater utilization of agricultural biomass fuels, further improvement of power plant efficiencies and solutions to the ash related problems of certain problematic biomass [1]. The ash related problems like deposit formation, corrosion and erosion are usually responsible for malfunctioning of combustion systems. These problems may have implications

on heat transfer rates and hence decrease the efficiency of the boilers. The deposits may restrict the gas flow in the boilers, through deposit accumulation and thus give rise to mechanical damages that turn the boiler unmanageable. In the limit, these problems may imply the facility shut down for maintenance. In general it can be distinguished between two types of deposit formation: slagging and fouling. Slagging refers to the deposition taking place in the high temperature refractory sections of boilers where radiative heat transfer is dominant and occurs due to the presence of molten ashes. Fouling takes place in the convective heat transfer zones of the boiler and is due to the formation of ash deposits, while the gases cool down. Several complex mechanisms of interaction of gases

* Corresponding author. Tel.: þ351 21 092 4600; fax: þ351 21 716 6569. E-mail address: [email protected]eg.pt (P. Teixeira). 0961-9534/$ e see front matter ª 2012 Elsevier Ltd. All rights reserved. doi:10.1016/j.biombioe.2012.01.010

b i o m a s s a n d b i o e n e r g y 3 9 ( 2 0 1 2 ) 1 9 2 e2 0 3

and particles with the heat transfer surfaces and their deposits are responsible for the fouling effects, depending on the technology used [2]. In Fluidized Bed Combustors (FBC), in addition to the deposit formation on heat transfer surfaces, the agglomeration of bed material may occur. FBC may be considered as one of the most promising technologies for biomass firing and cofiring, because of its fuel flexibility and relatively low combustion temperatures. The tendency for agglomeration of some biomass ashes and sand from the bed, particularly for agricultural biomass, could be a serious problem. Bed agglomeration may decrease heat transfer and fluidization quality, and may lead to the total defluidization of the bed, resulting in unscheduled plant shut down [3e5]. Some mechanisms are proposed in literature to explain the agglomeration phenomena that occurs in FBC due to complex interaction of ashes with bed material, which is usually sand [3,6]. The agglomeration could start with sintering, a phenomenon where loosely attached particles become denser, through the adhesion of small ash particles on the sand particles surface. This adhesion is promoted by the reaction or collision between melted ash particles (salt and silicate melts) with low fusion temperatures points and bed material that starts to became coated by sticky ash layer. With the continuous deposition of the ash in bed particles surface, and depending on the melting behaviour of the coatings, it can occur partial melting and agglomeration of the bed particles [3,7]. Two possibilities are associated to the melting occurrence: (1) low viscous melts that may occur when salts of alkali (mixtures of Na and K sulphates, chlorides and carbonates) and salts of alkaline earth metals (mixtures of Ca and Mg sulphates, chlorides and carbonates) are present in the ash; (2) highly viscous melts (or viscous flow sintering) that may occur when silicon is present in the ash; silicates can melt and form a glassy phase that does not crystallize when the temperature decreases back below the melting range of the silicate [7,8]. Over several years many mathematical indexes have been developed to predict coal slagging and fouling behaviour during combustion. Usually, they take into account the ash forming elements present in the fuel, which have been identified as being important in promoting those phenomenons. Often, they are relatively simple ratios of the concentration of several elements that form the fuel ash matrix [9]. More simple considerations relating elements concentration with ash melting tendency, during biomass combustion, can be also found in the literature. For example, Obernberger et al. [10] reported that the use of biomass with Ca concentrations below 15 wt% (d.b.) and K concentrations above 7 wt% (d.b.) in the ash, present a higher tendency for slagging problems. In fact, the elemental composition of biomass differs from that of coal, although biomass is a general designation for very heterogeneous and diverse biologic origin materials that may present quite different compositions. Vegetal biomass generally presents higher levels of K and P; sometimes Ca and Si are equally higher, but the levels of Al, Fe and Ti are usually lower than in common coals. In addition, biomass has a particular feature in comparison with coals; the alkali and alkali-earth elements are often in ionic forms or organically bounded rather than associated with minerals. As

193

a consequence, during the combustion process these elements, that are more volatile in ionic and organic forms, may easily vaporize and after condensation form deposits on the convective sections of the boilers, interact with the bed material originating agglomeration problems, or even form deposits on refractory sections of boiler [11]. The utilization of indexes based on ash composition to predict slagging and fouling problems appear to be an easy and expedite methodology and it should be desirable that these indexes could be equally valid to biomass combustion. Their application still could make some sense when very small amounts of biomass replace coal as a fuel, however, for higher levels of coal replacement with biomass, or when biomass is burned alone, ash problems prediction using those indices could be unrealistic. Some limitations about their application to biomass combustion are indicated below: (i) First of all the reactivity of ash forming elements present in biomass appears to be a relevant factor for the ash related problems which is not considered. The indices might be improved considering the reactivity of the elements depending on the fuel. (ii) One of the most common indices that correlates Base-toAcid oxides, RB/A, in ash is that of Eq. (1) [12]: RB=A ¼

ðFe2 O3 þ CaO þ MgO þ K2 O þ Na2 OÞ ðSiO2 þ TiO2 þ Al2 O3 Þ

(1)

RB/A considers that the alkali and alkali-earth oxides have the same role in the melting formation relatively to SiO2 content (in biomass Ti and Al contents are not significant), i.e., the basic compounds, B, are assumed to decrease the melting temperature, while the acidic ones, A, increase it. However the K2OeCaOeSiO2 ternary equilibrium phase diagram (Fig. 1) [13] shows that an increase of CaO relatively to K2O allows an increase of the fusion temperature point of the melts. This happens because Ca may affect the equilibrium reactions, i.e., the calcium silicates formation or potassium-calcium-silicates could be favoured relatively to potassium silicates that melt at lower temperatures [14,15]. (iii) A simplification of RB/A index is presented in Eq. (2) [12]: RB=A Simp: ¼

ðFe2 O3 þ CaO þ MgOÞ ðSiO2 þ Al2 O3 Þ

(2)

However, this index is not likely to be appropriated for biomass because it does not consider potassium which is one of the main responsible for slagging and fouling problems. For coals, prediction of slagging potential through indices of Eqs. (1) and (2) was evaluated by comparison with the ash fusion temperatures, and it was verified that the correlation was not linear. It was found that slagging was higher for RB=A z0:75 (hemispherical temperature under 1200  C), was lower when RB/A increased from 0.75 to 2 or when RB/A decreased below 0.75 (for RB/A ¼ 0.15 the hemispherical temperature exceeded 1600  C) [12]. (iv) The application of indexes based on sulphur content, RS, to biomass [12] may also produce incoherent predictions,

194

b i o m a s s a n d b i o e n e r g y 3 9 ( 2 0 1 2 ) 1 9 2 e2 0 3

Fig. 1 e Ternary phase diagram of CaOeK2OeSi2O (temperature unit in  C) [13].

in part due the lower levels of sulphur concentrations in biomass comparing with coal: Rs ¼ RB=A  Sd

Sd is the percentage of sulphur in dry fuel



(3) For coals the slagging potential was found to be low when RS < 0:6, medium when RS ¼ 0:6  2:0, high when RS ¼ 2:0  2:6, and extremely high for RS > 2:6 [12].Utilization of Cl content (or both Cl and S) could be a better choice as in biomass Cl is usually present in higher quantities and plays an important role in the vaporization of alkali elements, namely potassium, which is one of the main agents of slagging and fouling problems. (v) The fouling index, Fu, which is based on Base-to-Acid ratio, according to Eq. (4), could be more adequate for biomass because it gives more relevance to the alkaline elements which are the main agents of fouling. Nevertheless, an improvement could be achieved if Cl and S contents were included because of their influence on the vaporization of alkali elements. Fu ¼ RB=A  ðNa2 O þ K2 OÞ

(4)

The fouling potential was found to be low when Fu  0:6, medium for 0:6 < Fu  1:6, high for1:6 < Fu  40, and extremely high with tendency to deposits sintering when Fu > 40 [12].

An alternative to those composition indices is the ash fusibility index (AFI) which has a close correspondence to the real ash melting behaviour of fuels. The AFI index, presented in Eq. (5), is based on the ash fusion temperatures which can be determined under reducing or oxidizing conditions according with standardized methods for coals or biofuels (see Table 1). IDT is the initial deformation temperature ( C) and HT is the hemispherical temperature ( C). AFI ¼

ð4  IDT þ HTÞ 5

(5)

However, previous studies performed with coal showed that, quite often, different initial deformation temperature may be obtained, depending on the use of reducing or oxidizing conditions, i.e., the initial deformation temperature is dependent on the atmosphere, which could origin some inaccuracy about the melting tendency of ashes [16]. The slagging potential was found to be severe for AFI < 1149  C, high for AFI between 1149  C and 1232  C, medium for AFI between 1232  C and 1343  C and low for AFI > 1343  C [9]. Some literature refers that slagging potential is severe for values of AFI < 1052  C, probably accounting for differences due to the oxidant and reducing atmospheres [16]. Among the indices described in the literature to predict the ash related problems due to melts formation, probably, the ash fusibility index is one of the most promissory for biomass. The AFI index does not explain the phenomenon

195

b i o m a s s a n d b i o e n e r g y 3 9 ( 2 0 1 2 ) 1 9 2 e2 0 3

Table 1 e Proximate analysis, heating values and ash fusibility of fuels.

Moisture (a.r., wt%) Ash (d.b., wt%) Volatile matter (d.b., wt%) Fixed carbon (d.b., wt%) Low heating value (d.b., MJ/kg)

Standard method

Polish coal

Colombian coal

Straw pellets

Olive cake

Wood pellets

ASTM D 3173 or CEN/TS 14774e3 ASTM D 3174 or CEN/TS 14775 ISO 562 or CEN/TS 15148 ASTM D 3172 ASTM D 5865 or CEN/TS 14918

2.1 6.2 32.2 61.6 28

9.3 9.2 37.5 53.3 27

10.6 5.8 76.6 17.6 17

7.9 4.9 76.7 18.4 19

8.4 0.4 86.2 13.4 19

1223

1202

819

751

1238

1233 1251 1284

1358 1397 1443

1014 1167 1238

830 1367 1386

1265 1282 1291

Ash fusibility temperatures ( C) (Oxidant conditions) Initial Deformation Temperature (IDT) Softening temperature (ST) ASTM D 1857 or CEN/TS 15370e1 Hemispherical temperature (HT) Fluid temperature (FT) a.r. e as receive; d.b. e dry basis.

associated with the deposits formation and agglomeration. However, it is a simple and expedite approach to predict the occurrence of ash related problems. Whereas the ash composition indexes are merely correlations of independent parameters ignoring any possible synergies between them, the ash fusibility index is based on the experimental thermal behaviour of the ash samples, allowing existent synergies that influence fusibility temperatures to be incorporated. Nevertheless, it is important to remember that the accuracy of AFI prediction is also limited because of the differences inherent to the ash production in laboratory to perform the ash fusibility analysis and the ash produced in the combustion installation, such as: gradients of temperature; changes on atmosphere conditions; ash particles interactions and ash compounds segregation due to differences of particles size and high temperature volatilization. Other more laborious approaches have been applied to the study of ash related problems. Recent studies involving compression strength tests [7], chemical equilibrium model calculations [10], and benchescale fluidized bed combustion tests [3,5] have been undertaken to better understand problems associated with biomass ash. Nevertheless, no definitive tools are yet available to completely predict and prevent ash related problems, as they depend not solely on the fuel characteristics, but also on the combustion technology applied and operational parameters used. In this work it was aimed to understand if fusibility indices could be successfully applied to the combustion of coals and different types of biomass, as well as to different blends of them, by comparison with experimental data obtained from experimental combustion tests. Ashes obtained during the combustion tests, carried out in a pilot FBC, were analysed by visual inspection and when some type of ash agglutination was observed a more detailed analysis was performed by scanning electronic microscopy e energy dispersive spectroscopy (SEM/EDS). For coals, both the ash composition indices and the fusibility index were applied to evaluate if different predictions were obtained. The accumulation of particles on a deposition probe during biomass combustion and co-combustion was also evaluated. The mineralogical composition of particles collected in the probe was studied by X-Ray diffraction (XRD) and correlated with the fouling tendency.

2.

Experimental section

2.1.

The fuels used

Three biomasses of different types were studied, namely; Straw Pellets (SP), Olive Cake (OC) and Wood Pellets (WP). According to CEN/TS 14961 standard, they may be included in different subgroups; herbaceous biomass, fruit biomass and woody biomass, respectively. Two bituminous coals were used, Polish coal (PC) and Colombian coal (CC). The physicalechemical properties of the used fuels are presented in Tables 1 and 2.

2.2.

Combustion conditions

Combustion of coal or biomass and co-combustion tests of coal/biomass blends with 5, 15 and 25% (w/w) of biomass, were carried out on a pilot scale fluidised bed combustor (FBC) for periods of about 3e8 h, depending on the stability conditions of the FBC system. The main goal was to perform trial combustion tests to infer the best conditions for full combustion and for evaluation of the behaviour of biomass ashes during combustion and co-combustion, and characterize the ashes produced. The combustor is made of refractory steel and is well insulated. The trials with SP were performed on the old FBC with a square cross section of 0.09 m2 and 5 m of height. The other trials were carried out on the recently installed FBC with 6 m of height and with a square cross section area of 0.12 m2. The fuel was supplied under gravity to the top of the bed zone with a variable speed screw feeder. The combustion air was divided between the fluidising air and the secondary air, supplied to the freeboard zone to achieve complete combustion of volatiles and control NOx formation. The bed material used was silica sand with 0.37 mm mean diameter. The bed temperatures of each test were adjusted considering its homogeneity during the combustion runs, i.e., it was defined that the difference of measured temperatures between the thermocouples located on the bed zone should be less than 10  C. If gradient temperatures were found to be higher this was interpreted as a sign that the fluidization of bed was not in good conditions, and probably some aggregation was taking place. At the same

196

b i o m a s s a n d b i o e n e r g y 3 9 ( 2 0 1 2 ) 1 9 2 e2 0 3

Table 2 e Elemental analysis of fuels. Elemental analysis (d.b., wt%) C H N S Cl P Al Fe Ca K Na Mg Si

Standard method

Polish coal

Colombian coal

Straw pellets

Olive cake

Wood pellets

70.98 4.89 1.22 0.51 0.255 0.017 0.694 0.307 0.513 0.104 0.045 0.222 1.011

66.23 5.94 1.39 0.65 0.07 0.003 1.108 0.453 0.171 0.208 0.011 0.132 2.674

46.67 6.97 0.73 0.14 0.270 0.132 0.012 0.010 0.359 1.306 0.029 0.077 1.258

50.60 6.95 1.09 0.11 0.337 0.248 0.011 0.030 0.291 2.109 0.054 0.085 0.059

49.62 6.87 < 0.2 <0.06 < 0.001 0.008 0.006 0.005 0.083 0.040 0.002 0.020 0.035

ASTM D 5373 or CEN/TS 15104 ASTM D 4239 ASTM D 2361 ASTM 2795

ASTM 3682

time, the variation of pressure in different locations of the installation was evaluated. In the case of PC/SP runs lower temperatures were used to prevent possible fluidization problems given the known problems of herbaceous biomass. A water cooled deposition probe was located at the top of the freeboard zone (gas temperature ranged 600  Ce760  C, depending on the test run) to collect particles that could form deposits on cool surfaces of heat exchangers. The operating conditions are given in Table 3.

3.

Results and discussion

During the combustion trials on the pilot FBC the agglomeration phenomena (that could result in slagging in longer period tests) was noticed during the run with 25% SP and was especially evident during tests of 100% SP where agglomerated particles with diameters up to 5 cm were observed, which naturally influenced the maximum fluidised bed temperature used (about 700  C for 100% SP combustion test) and duration of the test. The agglomerates were probably formed as a consequence of the high contents of SiO2 and K2O in SP ash.

It was verified that the SP bulk (in an ash basis), contained about 46% of SiO2 and about 27% of K2O. CaO that according the CaOeSi2OeK2O equilibrium phase diagram (Fig. 1) generally allow an increase of the melting temperature, only represent about of 9% of ash content. The bulk content of the bottom ashes (including agglomerates) in the FBC differs considerably from the bulk content of the fuel ash. This happen due the presence of sand particles on the bed zone of FBC and due the partitioning of ash particles during the combustion related with the reactivity and melt behaviour of the elements. However, the use of the CaOeSi2OeK2O ternary diagram could be also useful to evaluate a priori the tendency to occur the formation of melts at low temperatures. For example, Scala and Chirone [17] observe that when the initial ash composition belongs to the region of the ternary phase diagram where the low melting point eutectics appear, due the low content of Ca on ashes and high Si content, the ash melting may already begin before the ash structure comes in contact with the sand particle. The authors refer that similar results were reported by Salour et al. [18] during the comparison of rice straw and wood wastes combustion behaviour, it was observed that straw had

Table 3 e Combustion conditions at the fluidised bed pilot. Fuel blend

Feed rate (kg/h) Energy input (MJ/h) Bed temperature ( C) Max. freeboard temperature ( C) Top freeboard temperature ( C) Bed gas velocity (m/s) Freeboard gas velocity (m/s) Excess air (%) Secondary air (%) Bed height (m)

Polish Colombian coal (%) coal (%)

Straw pellets þ Polish coal (%)

Olive cake þ Colombian coal (%)

Wood pellets þ Colombian coal (%)

100

100

5

15

25

100

5

15

25

100

5

15

25

100

9.56 266 818 819

11.4 280 850 865

9.77 265 813 835

10.4 269 813 839

10.6 260 769 862

15.7 233 701 960

12.4 318 845 867

12.9 321 840 867

13.9 334 831 878

20.3 353 766 974

12.4 300 843 874

13.4 313 842 871

13.8 313 841 900

18.3 316 831 968

622

662

630

644

643

728

604

663

682

760

653

658

726

766

1.1 1.3

0.66 0.87

1.0 1.2

1.0 1.3

0.9 1.2

0.7 1.1

0.82 0.98

0.81 1.00

0.84 1.06

1.03 1.47

0.71 0.95

0.71 1.00

0.70 1.00

0.79 1.16

35 18 0.17

36 23 0.50

31 20 0.17

36 21 0.18

31 23 0.17

52 35 0.16

40 25 0.50

35 25 0.50

35 25 0.50

45 25 0.50

37 25 0.50

36 28 0.50

36 30 0.50

38 33 0.50

197

b i o m a s s a n d b i o e n e r g y 3 9 ( 2 0 1 2 ) 1 9 2 e2 0 3

3.1.

Evaluation of slagging tendency

3.1.1.

Fusibility index

As can be observed in Fig. 2 the values of the fusibility index obtained for each fuel and their blends are in agreement with the experimental results. According to the fusibility index SP and OC (with AFI of 888  C and 874  C, respectively) are predict to behave as problematic fuels. The AFI values fall in the range of severe slagging for both biofuels if burned alone. In the SP trial tests, 100% and 25%, it was observed the tendency for formation of agglomerates, and for OC it was verified that the diameter of the sand particles used as bed material increased (a sign of sintering). For WP no such problem was predicted by the AFI value, as was observed during its mono-combustion. The AFI values for the blends tested, 5, 15 and 25% of biofuel suggest that it is possible to control the tendency for sintering of the problematic fuels through their blending with coals. However, a higher tendency of SP for ash melting relatively to OC was evidenced by the low fusibility index of 25% SP blend with PC, with AFI of 1156  C being close to the limit of severe slagging. This result is coherent with the experimental evidences. It was quite interesting to verify that in spite of the lower Initial Deformation Temperature (IDT) of OC than that of SP (Table 1), the AFI of both biomasses was found to be similar because the Hemispherical Temperature (HT) of OC was 200  C higher than HT of SP. In fact, previous studies showed that peculiar problems may arise for biomass with high contents of silica as is the case of the herbaceous biomass (SP) tested. In those cases the melting process may start at temperatures lower than the first deformation temperature observed during the fusibility test. This fact may be related to an artefact occurring during the experimental fusibility determination in which the high levels of silica may form a rigid cage like structure in the surface of the pyramidal pellet ash during its oven heating [20]. The main conclusion, comparing the combustion and co-combustion tests, is that it is possible to improve fuel blend composition by paying attention to the determination of the ash fusibility temperatures. One possibility to decrease the slagging potential could be blending biofuels that present low fusibility temperatures, such an SP and OC, with less problematic fuels, like CC and WP.

1350

Medium

Ash Fusibility Indice (ºC)

tendency to agglomerate, which could be related with the high silicon content of straw and low calcium content comparatively with the wood wastes. A common difficulty associated to the ternary diagrams utilization is related with the fact that the ashes are constituted by several oxides, which are not considered in the diagram. According to the literature, to obtain accurate predictions of the melting points, the sum of the oxides considered in the ternary diagram, in this case CaO, SiO2 and K2O, should be higher than 90% [3]. Beside this, as mentioned before for the indexes approach, parameters related with the combustion conditions and fluidizing air in FBC are not contemplated in the results obtained through the use of phase diagrams. The diagram presented in Fig. 1 identifies the initial melting temperatures as a function of oxide composition and at the same time the expectable mineralogical composition. Interesting for the present study is the indication that around 750  C some melt can be expectable due to the presence of a potassium silicate (K2O.4SiO2). This tendency of herbaceous biomass to form potassium silicates with low melting temperatures during combustion is well described in the literature [19]. During OC pure combustion the bed temperature was maintained below 770  C to avoid agglomeration problems. Although no agglomeration was detected under these conditions it was verified that the diameter of sand particles from the fluidised bed increased, which suggests that some ash sintering occurred. This may be explained in part by the OC bulk composition in an ash base, i.e., the ash contained about 52% of K2O, about 3% of SiO2 and 8% of CaO on its constitution. The formation of potassium silicate compounds by the ash fuel constituents, as it occurred for SP, appears not to be so significant during OC combustion because it is limited by the low availability of silicon in the fuel ash. For co-combustion trials of OC this effect was not so evident. In the case of WP combustion trials bed agglomeration problems were not detected, as was expected given the low ash content (about 0.4%) and its oxide constitution, namely; 29% of CaO, 12% of K2O and 19% of SiO2. The experimental difficulties observed during SP and OC mono-combustion trials are consistent with the Ca and K limits for slagging problems found in the literature (less than 15% of Ca and more than 7% of K) [10]. In fact the SP ash bulk contains 6% of Ca and 23% of K and the OC ash exhibits 6% of Ca and 43% of K. For WP the Ca content was about 21% and K attained 10%, which in the case of K exceeds the 7% safe limit. However, probably due the low ash content, calcium content above 15% and the short duration of the combustion tests performed, no problems were identified for WP trials. The experimental co-combustion tests of different biomass studied showed that the tendency for ash sinterization and, in worst case agglomeration, of straw and olive cake, could be mitigated through their blending with coals. However, in the case of SP the 25% share with PC appeared not to be effective enough as the temperature of the fluidised bed had to be maintained below 800  C to prevent severe agglomeration, because still some sintering effects were detected.

1250

High 1150

1050

Severe 950

850 100% Coal

5% Biofuel

SP + PC

15% Biofuel

25% Biofuel

OC + CC

100 % Biofuel

WP + CC

Fig. 2 e Slagging potential (AFI) of different fuels and blends.

198

b i o m a s s a n d b i o e n e r g y 3 9 ( 2 0 1 2 ) 1 9 2 e2 0 3

Table 4 e Evaluation of slagging and fouling tendency based on different indexes. Fuel

Eq. (1)

Polish coal Colombian coal

3.1.2.

Medium (0.5) Low (0.2)

Rb=a Simp Eq. (2)

RS Eq. (3)

Fu Eq. (4)

AFI Eq. (5)

Medium (0.4) Low (0.2)

Low (0.2) Low (0.1)

Medium (1.4) Low (0.6)

High (1229  C) Medium (1241  C)

Ash composition indexes

Some available indexes developed to predict coal slagging and fouling behaviour during combustion were applied to the coals tested and the results are summarized on Table 4. Due to the limitations of ash composition indexes application to biomass, as was mentioned before, they were applied only to the used coals. As can be observed in Table 4, in general, the calculated indexes give indication that Polish coal presents higher tendency for slagging and fouling than the Colombian coal. Exception was observed for RS that besides the ash oxides, also considers the sulphur content. The AFI values are similar, as can be seen in Fig. 2 which means that in spite of the different classification of slagging degree based in the literature, both values are close to the limits, hence no clear differences in slagging tendency should be drawn. The different values obtained for the other indexes may be due to PC higher elements content in alkali-earth and lower in silicon and aluminium contents comparing with CC. This is in line with the idea that for coals the tendency to form deposits is usually associated to the higher quantity of basic oxides comparing to the acidic ones. However, for the range of temperatures used during FBC coal trials, between 818  C and 850  C, no differences were observed in ash effects between the two coals.

3.1.3. Ash characterization using scanning electronic microscopy e energy dispersive spectroscopy The scanning electronic microscopy e energy dispersive spectroscopy (SEM/EDS) analyses were applied to the bed ash samples obtained during mono-combustion of SP and OC because in these combustion tests some agglutination/sintering effects were observed experimentally, confirming prediction using the fusibility index. The agglomerates founded on the bed ashes were mounted in an epoxy resin which took at least a day to achieve complete hardening. Cross section areas were cut and polished with different polishing papers, until 1 mm grain size. This type of preparation allows to observe the build up of layers around the particles and the morphology of the agglomerates, and to identify the major elements that may be responsible for the melting effects. It was used a high resolution Philips XL 30 FEG/EDAX NX scanning electron microscope. Samples were coated with a thin golden layer using an ion sputter JLC 1100, in order to obtain the conductivity needed for SEM observation. Software included in the equipment allowed data acquisition for semi-quantitative analysis and to perform elements mapping. Fig. 3 shows a SEM picture of a bed ash mounting containing individualized particles and agglomerates from combustion of 100% SP. Some individualized sand grains could be observed but many of them were glued each other. In Fig. 4 the elements mapping of the area observed in Fig. 3 is

presented. Table 5 resumes the semi-quantitative analysis that was performed in specific zones of the SEM image presented in Figs. 3 and 5. Fig. 4 shows that the distribution of Si and O match each other, corresponding to the original grain sand forms of the silica sand used as bed material in the FBC tests, as shown in areas A1, A3, A6 and A5, for which the semi-quantitative analysis detected only Si and O. In the aggregated areas, there are distinct clear grey zones in which the sand particles appear to be embedded, as shown in A4 and A8 areas and on the left of A3. These zones, acting like agglutination glue of the sand, present different composition from the silica grains. Although Si and O are present in these areas, higher concentrations of K, some Ca and vestiges of Mg are visible in the maps and also on the semi-quantitative analysis. This means that probably potassium silicates and also potassiumecalciumesilicates were formed. As mentioned before and evidenced in K2OeCaOeSiO2 diagram of Fig. 1, usually the potassium silicates present lower melting points (about 700  C) than the potassiumecalciumesilicates (about 900  C) or calcium silicates (higher than 1300  C), which suggests that the agglutination mass may have originated from the melts of potassium silicates. In spite of the temperature in the bed zone was maintained relatively constant at about 700  C, during SP combustion, hot spots were observed in the freeboard temperatures reaching values up to 960  C (Table 3), which means that even the potassiumecalciumesilicates that could be present also contributed to the melt formation. During the combustion, if sticky melts of potassiumecalciumesilicates contact with projected bed and ash particles, some adhesion between the melts and these particles could happen. When these particles return to the bed zone and the temperature decreases, the melts solidify which could contribute to the agglomeration of bottom ashes.

Fig. 3 e SEM image of SP agglomerated bed ashes.

199

b i o m a s s a n d b i o e n e r g y 3 9 ( 2 0 1 2 ) 1 9 2 e2 0 3

Fig. 4 e Elements mapping of SEM image of SP agglomerated bed ashes.

Simultaneously, during the combustion on bed zone, the fuel particles surface (pellets), which are burning at higher temperatures than the given average bed temperature can also contribute to the agglomeration.

K and some Ca are also visible in the grain contours of some sand particles, e.g., zones A1, A3 and A6 suggesting that probably some potassiumecalciumesilicates adhere to the sand surfaces.

Table 5 e Semi-quantitative analysis from specific zones of SEM images. Elements (%)

A2

Al Ca Cl Fe K Mg Na P O S Si Ti

17.6

A4

A5

A7

A8

37.5

5.8

B1

B2

B3

4.0 5.2

4.9 20.2 3.2

13.3 1.1

34.8

37.1

41.5

19.3

43.3

58.5

40.7 18.1 3.7

B4

B5

4.6 6.6 0.2 5.9 18.5 5.5

B6

14.7 1.1

18.1 0.8

8.58

11.2

6.7

35.0

35.4

40.50

33.4

14.1 32.6

43.0

42.8

43.4

41.6

50.92

55.4

12.0

57.0

50.4

200

b i o m a s s a n d b i o e n e r g y 3 9 ( 2 0 1 2 ) 1 9 2 e2 0 3

Fig. 5 e SEM images of OC bed ashes.

Other particles like A2 and A7 present a different composition and do not appear to contribute to the agglomeration or slagging phenomena. A2 is an isolated particle containing low amounts of O, K, Al, Fe and Mg and lower Si content than other sand particles, hence it may be an ash particle or extraneous minerals present in the straw. A7 corresponds to a particle containing high quantities of Ca, S and O and low Si; hence it may correspond to calcium sulphate. In the case of OC bottom ashes the SEM image presented in Fig. 5 reveals that particles were kept individualized without major agglomeration, but their peripheral zones show a different morphology and some of them are connected by small points and thin bridges. EDS of the necks formed between particles or isolated particle surfaces (B1, B2, B3 and B6) showed the presence of potassium in addition to Si and O, although in some restrict zones low quantities of Al and Mg may have been incorporated (B1). It appears also that particle surface tend to stick individual ash particles, as in the case of B4, containing a more complex composition (O, K, P, Si, Mg, Fe, Ca, Al and Cl). Elements mapping in Fig. 6 confirm that silica base particles are covered by uniform superficial layers containing variable concentrations of K.

From the observations of the two types of bed ash material produced in the SP and OC combustion tests, it appears that two different phenomena took place. In the case of SP, the high contents of Si and K of the biomass ash probably allowed the formation of potassium silicates that melts at low temperatures and in hottest points of the freeboard also potassiumecalciumesilicates may have contributed to the melting pot. The formed melts covered the sand particles and formed gluing ashes that agglomerated the sand particles resulting in hard compact structures, without major internal interactions with the sand particles. In the case of OC, the higher proportion of K and lower content of Si in their ash may have resulted in potassium interactions with the silica sand particles, forming a progressive intrusive potassium silicate layer from the surface core towards the inner particle. This may have contributed to enlarge the size of sand particles, sticking some of them together through thin bridges or necks. The smoothest surface appearance of particles, in comparison with those of the SP test, also seems to indicate that surface melting may have occurred enabling the interaction of sand particles covered by the melt layer with the fuel ash

Fig. 6 e Elements mapping of SEM image of OC bed ashes.

201

b i o m a s s a n d b i o e n e r g y 3 9 ( 2 0 1 2 ) 1 9 2 e2 0 3

Olive Cake

3.2.

Woodpellets

Evaluation of fouling tendency

60

3.2.1Ratio of particles deposition

Deposit (g/m2h)

50

The ratio of accumulation of particles (g/m2h) on the deposition probe when biomass replaced coal is presented in Fig. 7. Comparison was made between OC and WP runs because they were performed in the same pilot FBC and used the same coal, CC, the one that was predict to be less problematic, to evaluate the influence of co-firing a good quality biomass and a more problematic one. It is possible to verify that in the case of WP trials the particles deposition did not varied significantly (10e20 g/m2h). In the case of OC the deposited quantity was much higher when 100% biofuel was fired than when cofiring with coal was performed. Apparently, for 5% and 15% biofuel the ratios of deposition were dominated by particles originated from coal as it was observed that the ratio of deposition was similar for the co-combustion with OC and WP. However, when the replacement of coal with biomass attained 25%, the different influence of biomass became notorious. As ash level of OC was almost half than that of the coal, the high deposition rate can only be attributed to the ash fouling nature of OC.

40 30 20 10 0

5% Biofuel

15% Biofuel

25% Biofuel

100% Biofuel

Fig. 7 e Ratio of accumulation of particles on a deposition probe located at pilot fluidized bed of LNEG.

particles. Brus et al. [6] also pointed out that, for long combustion periods of olive residue, an inner and outer layer was formed on the quartz silica sand; the inner layer containing mainly K and Si, and minor quantities of Ca and Mg, meanwhile the outer layer was more complex and similar to the fuel ash composition. In our study evidences of this tendency were observed when firing pure OC and it is believed that similar effects could have occurred if experiments lasted for a longer period. From the observations made, one could suggest that a possibility to prevent this type of phenomena is to use different bed materials such as olivine or others from the alumina-silicate group, containing e.g. Mg, Fe and Al in their composition, to prevent reactions of potassium with silica from the sand.

3.2.2. Characterization of particles deposited using X-Ray diffraction The deposits collected on the probe during OC and WP monocombustion were analysed by XRD to identify main differences of composition and understand the different fouling tendency of the studied biofuels.

Intensity (Counts)

1000

Olive Cake Deposit Probe

A

A

A

750

A A

D

A

500 A B C B

250

B

E

B

A

A

A

A F C

F

A

C

D

0

0

10

20

30

1000

Intensity (Counts)

Wood Pellets Deposit Probe

40 50 Two-Theta (deg)

60

70

80

A

A

A

750

A

A A

500 A G

250

H

H I GG

A A

A

A

A GI

I

G A

0

0

10

20

30

40 Two-Theta (deg)

50

60

70

80

Fig. 8 e XRD of particles collected on the deposit probe during OC and WP mono-combustion. A-SiO2, B-CaSO4.2H2O, CK2SO4, D-KCl, E-K3PO4, F-K2S2O3, G-Fe2O3, H-CaSO4, I-KOH.

202

b i o m a s s a n d b i o e n e r g y 3 9 ( 2 0 1 2 ) 1 9 2 e2 0 3

As can be observed in Fig. 8 the bed material (SiO2) was the dominant compound of the probe collected particles. As SiO2 is not volatile under FBC operating conditions, this means that a significant degree of elutriation occurred during the combustion tests, as it was mentioned in previous published work [21]. In the case of OC mono-combustion, among other potassium compounds, K2S2O3, K2SO4 and KCl were identified on the probe deposits. The sulphates may be originated by sulphation reactions given the high Cl content of OC, through the “quenched equilibrium” which may give rise to homogeneous nucleation of K2SO4 followed by KCl condensation [22] These salts are referred in the literature as the main responsible by fouling and corrosion problems [23], hence it is expectable a higher degree of fouling during OC combustion. K3PO4 was also identified in deposits of OC monocombustion. In spite of the phosphor reactions with other ash constituents are not well established yet, previous studies [24] evidenced that usually it is associated with the deposition on convective zones, where the fouling is more pronounced. The particles collected on the probe during WP monocombustion did not show low melting salts. Vestiges of KOH indicate that a slight volatilization of K occurred even for WP, but in the absence of S and Cl only the hydroxide was formed, hence it was not expected the occurrence of significant fouling problems during wood pellets combustion. Identification of Fe2O3 may indicate some corrosion effects, although no significant importance should be given to corrosion effects on these short combustion trials and there is also the possibility that some particles from FBC internal corrosion may have deposited on the probe.

4.

Conclusions

Some conclusions could be drawn from the present work. The woody biomass can be successfully used as biofuel without significant slagging and fouling problems. Use of herbaceous and fruit biomass may lead to slagging and fouling, however this tendencies can be mitigated through co-combustion with appropriate coals, under well controlled temperatures. These tendencies were anticipated a priori by the calculated fusibility index and confirmed experimentally during combustion tests. The fusibility index revealed to be a useful methodology to predict slagging or agglomeration problems for coals, biomass and inclusively blends of both fuels. From the results obtained it appears that it is the presence of enough quantities of Si and K in the biomass that favours the formation of highly sticking ashes that are responsible for ash sintering, which can be followed by agglomeration, such as in the case of SP. When great amounts of Si and K are available in the biomass ashes, the formation of potassium silicates is thermodynamically favoured, and the formation of melt can be observed even at low combustion temperatures. However, when Ca is present on biomass ash, potassiumecalciumesilicates are also formed, increasing the fusion temperature of the ashes. Nevertheless, even they can contribute to the agglomeration due to the occurrence of the hot spots during biomass combustion in the freeboard. During the combustion,

potassiumecalciumesilicates melts could contact and adhere to the projected bed and ash particles, and when these particles return to the bed zone the melt solidify and could contribute to the agglomeration. The agglomeration could also be promoted by these particles contact with the fuel particles surface (pellets), which are burning at higher temperatures than the given average bed temperature. If the availability of K in ash biomass is very higher comparing with Si content, interactions between the surfaces of bed silica particles are more prone to occur and K may be incorporated in the sand matrix surface, increasing their stickiness. This implies formation of covering layers on the sand particles surface, enlarging their diameter and given their sticky properties some bridges may build up between particles. The ratio of deposited particles on a deposit probe was found to be higher for OC than for WP mono-combustion, which was expectable because the different ash content of both biofuels. Moreover, the identification of KCl and K2SO4 on the OC deposited particles showed that fouling tendency is highly probable for OC combustion, due to its ash constitution. In terms of operating conditions, results indicate that the combustion of pure straw, or in shares attaining 25% with coal shall be done under well controlled temperatures below 700  C, avoiding formation of hot spots in the bed zone to assure good fluidizing conditions during longer run periods. For OC, continuous combustion may require the use of alternative materials as bed fluidizing agents instead of silica sand. Further investigation is being pursued for a better understanding of bed ash problems during biomass combustion in FBC.

Acknowledgements This work is part of COPOWER project: “Synergy Effects of CoProcessing of Biomass with Coal and Non-Toxic Wastes for Heat and Power Generation”. Financial support given by 6th Framework Programme is gratefully acknowledged. Financial support of a Doctoral grant (SFRH/BD/30076/2006) given by the Portuguese Foundation of Science and Technologies is also acknowledged.

references

[1] Obernberger I. TUE in http://alexandria.tue.nl/extra2/redes/ obernberger2005.pdf [accessed 14.09.07]. [2] Eubionet. Biomass co-firing e an efficient way to reduce green house gas emissions; 2003. [3] Ohman M, Nordin A, Skrifvars B, Backman R, Hupa M. Bed agglomeration characteristics during fluidized be combustion of biomass fuels. Energy Fuels 2000;14:169e78. [4] Llorente M, Garcıa J. Comparing methods for predicting the sintering of biomass ash in combustion. Fuel 2005;84: 1893e900. [5] Lundholm K, Nordin A, Ohman M, Bostrom D. Reduced bed agglomeration by co-combustion biomass with peat fuels in a fluidized bed. Energy Fuels 2005;19:2273e8.

b i o m a s s a n d b i o e n e r g y 3 9 ( 2 0 1 2 ) 1 9 2 e2 0 3

[6] Brus E, Ohman M, Nordin A. Mechanisms of bed agglomeration during fluidized-bed combustion of biomass fuels. Energy Fuels 2005;19:825e32. [7] Skrifvars B, Backman R, Hupa M. Characterization of the sintering tendency of ten biomass ashes in FBC conditions by a laboratory test and by phase equilibrium calculations. Fuel Process Technol 1998;56:55e67. [8] Paulrud S. Upgraded Biofuels e Effects of Processing, Handling Characteristics, Combustion and Ash Melting. Doctoral thesis, Swedish University of Agricultural Science, Umea, 2004. [9] Lopez CH, Unterberger S, Maier J, Hein K. Overview of actual methods for characterization of ash deposition. ECI conference on heat exchanger fouling and cleaning. Fundam Appl 2003;RP1:279e88. [10] Obernberger I, Brunner T, Barnthaler G. Chemical properties of solid biofuels e significance and impact. Biomass Bioenerg 2006;30:973e82. [11] Onderwater M, Blomquist J, Skrifvars B, Backman R, Hupa M. The prediction of behaviour of ashes from five different solid fuels in fluidised bed combustion. Fuel 2000; 79:1353e61. [12] Pronobis M. Evaluation of the influence of biomass cocombustion on boiler furnace slagging by means of fusibility correlations. Biomass Bioenerg 2005;28:375e83. [13] Slag atlas. 2nd ed. Verlag Stahleisen GmbH; 1995. [14] Risnes H, Fjellerup J, Henriksen U, Moilanen A, Norby P, Papadakis K, et al. Calcium addition in straw gasification. Fuel 2003;82:641e65. [15] Thy P, Lesher C, Jenkins B. Experimental determination of high-temperature elemental losses from biomass slag. Fuel 2000;79:693e700.

203

[16] Raask Erich. Mineral impurities in coal combustion. Behaviour, problems, and remedial measures. Hemisphere Publishing Corporation; 1985. [17] Scala F, Chirone R. An SEM/EDX study of bed agglomerates formed during fluidized bed combustion of three biomass fuels. Biomass Bioenerg 2008;32:252e66. [18] Salour D, Jenkins B, Vafaei M, Kayhanian M. Control of in-bed agglomeration by fuel blending in a pilot scale straw and wood fueled AFBC. Biomass Bioenerg 1993;4:117e33. [19] Geyter S, Ohman M, Bostrom D, Eriksson M, Nordin A. Effects of non-quartz minerals in natural bed sand on agglomeration characteristics during fluidized bed combustion of biomass fuels. Energy Fuels 2007;21:2663e8. [20] Natarajan E, Ohman M, Gabra M, Nordin A, Lliedahl Rao AN. Experimental determination of bed agglomeration tendencies of some common agricultural residues in fluidized bed combustion and gasification. Biomass Bioenerg 1998;15(2):163e9. [21] Teixeira P, Lopes H, Gulyurtlu I, Lapa N. Use of chemical fractionation to understand partitioning of biomass ash constituents during co-firing in fluidized bed combustion. Fuel; 2011. doi:10.1016/j.fuel.2011.07.020. [22] Christensen K, Stenholm M, Livbjerg H. The formation of submicron aerosol particles, HCl and SO2 in straw-fired boilers. J Aerosol Sci 1998;29(4):421e44. [23] Zheng Y, Jensen P, Jensen A, Sander B, Junker H. Ash transformation during co-firing coal and straw. Fuel 2007;86: 1008e20. [24] Piotrowska P, Zevenhoven M, Davidsson K, Hupa M, Amand L, Barisic V, et al. Fate of alkali metals and phosphorus of rapeseed cake in circulating fluidized bed boiler part 1: cocombustion with wood. Energy Fuels 2010;24:333e45.