Use of chemical fractionation to understand partitioning of biomass ash constituents during co-firing in fluidized bed combustion

Use of chemical fractionation to understand partitioning of biomass ash constituents during co-firing in fluidized bed combustion

Fuel 101 (2012) 215–227 Contents lists available at ScienceDirect Fuel journal homepage: www.elsevier.com/locate/fuel Use of chemical fractionation...

2MB Sizes 0 Downloads 82 Views

Fuel 101 (2012) 215–227

Contents lists available at ScienceDirect

Fuel journal homepage: www.elsevier.com/locate/fuel

Use of chemical fractionation to understand partitioning of biomass ash constituents during co-firing in fluidized bed combustion Paula Teixeira a,⇑, Helena Lopes a, Ibrahim Gulyurtlu a, Nuno Lapa b a b

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

a r t i c l e

i n f o

Article history: Received 29 September 2010 Received in revised form 2 May 2011 Accepted 13 July 2011 Available online 4 August 2011 Keywords: Chemical fractionation Partitioning Ash enrichment Biomass Co-firing

a b s t r a c t Three species of biomass origin (straw pellets, olive cake and wood pellets) and two coals from different countries (Coal Polish and Coal Colombian) have been studied to understand the fate of their ash forming matter during the combustion process and to investigate the influence of co-firing biomass with coal. Three different approaches to investigate the ash behaviour were employed: (1) chemical fractionation analysis to evaluate the association/reactivity of ash forming elements in the fuels as a prediction tool, (2) establishment of elements partitioning in ash streams produced in the combustion and co-combustion trials, and (3) evaluation of enrichment factors of elements in the ash streams. The chemical fractionation analysis was applied to all fuels used to evaluate how the association/reactivity of elements making up ash may influence their behaviour during combustion. Combustion tests were carried out on a pilot scale fluidized bed combustor (FBC). Four ash streams were obtained at different locations. The uncertainty of measurements was estimated allowing a critical evaluation of mass balances over the combustion system and the partitioning of elements in the ash streams. The enrichment factors of elements in the several ash streams were estimated, incorporating uncertainties associated with analytical measurements. Results obtained showed that for FBC the relation between the chemical fractionation and the experimental partitioning is strongly affected by elutriation of particles. The element enrichment factor estimated for each ash stream, using Al as a reference element, revealed better correlations with the elements reactivity obtained by chemical fractionation because it overcomes particles elutriation effects. Nevertheless, it was observed that the reactivity estimated by chemical fractionation could not be solely interpreted as tendency of the elements to volatilize on FBC system, as reaction in bed zone of boiler may also occur retaining reactive elements. Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction Since the 1990s, environmental and economic related issues mostly arising from the need to decrease greenhouse gas emissions and increasing fossil fuel prices encouraged the use of renewable energy sources. The EU Directive 2009/28/EC [1] on the promotion of energy from renewable sources set targets for the share of renewables in the energy production that are constantly being reviewed in the light of the growing concerns about the climate change. However, coal still contributes to nearly 40% of worldwide electricity generation and will maintain to have an important role in the global energy supply. One way to contribute to targets of renewable share and to maintain competiveness in the energy market is the co-combustion of biomass with coal in existing installations, which allows the use of biomass without major needs for retrofitting. Biomass co-combustion with coal represents a ⇑ Corresponding author. E-mail address: [email protected] (P. Teixeira). 0016-2361/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.fuel.2011.07.020

short term, low risk, low cost and sustainable reduction in net CO2 emissions, simultaneously reducing SOx and often NOx emissions and offering other several side benefits. Technical issues associated with co-combustion include fuel supply difficulties, security and additional handling and storage cost requirements, potential increase in corrosion of boiler tubes and decrease of overall efficiency, changes in the nature of ash deposit formation and unburned carbon and finally impact on potential ash reutilization [2]. The biomass concept can be more or less extensive, in accordance with the European Union legislation biomass is ‘‘. . .the biodegradable fraction of products, waste and residues from agriculture (including vegetal and animal substances), forestry and related industries including fisheries and aquaculture, as well as the biodegradable fraction of industrial and municipal waste’’ [1]. More restricted is the European Standard for solid biofuels definition, CEN/TS 14588 [3] which defines biomass as a ‘‘material of biological origin excluding material embedded in geological formation and transformed to fossil’’. The CEN/TC 335 responsible for the

216

P. Teixeira et al. / Fuel 101 (2012) 215–227

Nomenclature XOutput_BA ratio between ash mass ignited from the bottom ashes and mass of ash in the fuel input + sand input in FBC XOutput_1Cy ratio between ash mass ignited from the 1nd cyclone and mass of ash in the fuel input + sand input in FBC XOutput_2Cy ratio between ash mass ignited from the 2nd cyclone and mass of ash in the fuel input + sand input in FBC XOutput_PM ratio between the mass of particulate matter and mass of ash in the fuel input + sand input in FBC MInput_AshFuel + MInput_Sand mass of ash in the fuel input + sand input in FBC (ash basis, kg) MOutput_BA output of ash mass ignited in the bottom (750 °C ignited ash, kg) MOutput_1Cy output of ash mass ignited in the 1st cyclone (750 °C ignited ash, kg) MOutput_2Cy output of ash mass ignited in the 2nd cyclone (750 °C ignited ash, kg) MOutput_PM output of mass of particulate matter (kg)

development of standards for solid biofuels defines in CEN/TS 14961 [4] four main groups of solid biofuels, woody biomass, herbaceous biomass, fruit biomass, blends and mixtures. Their properties and composition are highly variable and differ from those of coals which may give rise to several problems. Comparing with coal, biomass has less carbon, more oxygen, higher volatiles content and some types present more silicon and calcium, more chlorine and potassium, and less aluminium, iron, titanium and sulphur. In addition to the different biomass fuel composition and properties that may influence the combustion process, the deposit formation and corrosion effects may particularly be relevant. The deposits can restrict the flow between the tubes of heat transfer surfaces of the boilers, through deposit accumulation which can cause mechanical damages or can turn the boilers unmanageable. Finally, such problems may imply the frequent shutdown of the plant for maintenance. Efforts need to be taken to minimize ash related problems that might occur when using different fuels. Investigating the proper blending of these fuels, benefits of ash interaction synergies may be achieve. In this study it was aimed to investigate the influence of reactivity of ash forming elements of biomass and coals on their behaviour/partitioning in the several ash streams for different combustion and co-combustion combinations. In fluidized bed combustion it frequently occurs elutriation of bed material which includes sand and ash, as well unburned fuel particles. This could lead to alterations in partitioning of elements, hence, enrichment factors were therefore evaluated using a reference element to overcome these effects. 1.1. Assessment of chemical association of elements in fuels The chemical fractionation employed is an analytical methodology based on the solubility of chemical species, when solvents with increasing aggressively are used in sequence. It allows evaluating the chemical association of elements that constitute the fuel, information that cannot be accessed by the traditional chemical characterization that reports only bulk composition. In the 1980s, Benson and Holm [5] compared the mode of incorporation of inorganic components in different low-rank coals through an extraction procedure using NH4Ac and HCl. Later, Miller et al. [6,7] adapted this methodology to biomass, including a leaching water step and changing the hot NH4Ac step to room

AshFuel(w) mass fraction of ash in the fuel (%) BA(w) mass fraction of ash in the ignited bottom ashes (%) 1Cy(w) or 2Cy(w) mass fraction of ash in the ignited 1st cyclone (or 2nd cyclone) ashes (%) C iOutput BA concentration of element i in the bottom ashes that output from FBC(mg/kg) C iOutput 1Cy concentration of element i in the 1st cyclone ashes that output from FBC (mg/kg) C iOutput 2Cy concentration of element i in the 2nd cyclone ashes that output from FBC (mg/kg) C iOutput PM concentration of element i in the particulate matter that output from FBC (mg/kg) C iIntput AshFuel concentration of element i in the ash in the fuel that enters in FBC (mg/kg) C iInput Sand concentration of element i in the sand that enters in FBC (mg/kg)

temperature. In recent years, this methodology has frequently been used and adapted by different research teams [8,9]. Typically, the more soluble compounds leached out by water are alkali sulphates, carbonates and chlorides. Recent studies showed that phosphates are also leached out in the water step [10]. NH4Ac generally leaches out the organically associated elements, especially Ca and Mg and some K and Na. Those removed by HCl are the carbonates and sulphates of alkaline-earth and other metals. Silicates and other refractory compounds remain in the insoluble fraction. This methodology may give an indication about the behaviour of the elements during high temperature processes, distinguishing the more reactive forms (ionic salts and bounded to the organic matrix) from those less reactive (minerals included and excluded of the fuel matrix). The elements with greater tendency to react, as it frequently occurs in biomass, can easily volatilize at high temperatures and subsequently condense and form deposits in the cooler parts like the convective pass of the boilers. However, especially in the case of alkaline metals, they may also interact with the bed material or bed ashes and contribute to agglomeration phenomenon, or even deposit on heat transfer surfaces and refractory walls of boilers. Inorganic elements that are present in biomass maybe classified in four classes [11,12]: soluble salts that are easily leached out by water; inorganic elements that are associated with the organic phases of biomass; inorganic and organometallic compounds present in the biomass matrix or exclude minerals, like sand or clay from biomass harvesting or atmospheric deposition on plants. In coals, the inorganic elements can be found as dissolved salts and other inorganic compounds incorporated within the organic matrix of the coal maceral, although they may also be present as distinct non-coal rocks minerals. The first two forms of mineral matter are usually prominent in brown coals, lignite and sub-bituminous coals, and contribute to deposit formation in low-rank coals. In bituminous coals and anthracites they are typically present in relatively low proportions [13]. 1.2. Mass balances and elements partitioning in ash streams For a consistent study of the behaviour of elements making up the ash in any thermal conversion system, mass balance should be performed in order to ensure that each stream is identified and that output flows match the inlet flows, or at least identify lack of closure and understand possible reasons.

P. Teixeira et al. / Fuel 101 (2012) 215–227

In fluidized bed combustor systems this is not so easily achieved, particularly on pilot-scale installation. As the scale of the installation increases it becomes more complex to control the quantitative recovery of all streams involving the combustor, because of losses due to poor recovery of ash deposits formed that may give rise to cross contamination when performing different tests on the same installation. Uncertainties of analytical measurements are associated with the mass of fuel input or ash and gas streams, ash and elements analysis, or even fuel heterogeneity, leading to inconsistency between laboratorial fuel ash and process ash, including the presence of unburned carbon or loss on ignition (LOI) and sand used as bed material in FBC. In FBC the elements of ash are very much diluted by the presence of sand, hence, in this study the mass balances for the combustion trials were performed using Eq. (1) that gives the recovery ratio of ash, RRAsh. As in the installation used for this study there are four output streams; bottom ash, BA, ash collected in a first cyclone, 1Cy, ash from a second cyclone, 2Cy, and particulate matter, PM, collected in the stack, they were incorporated in the calculation together with the sand used as fluidizing agent in the FBC. The list of symbols is listed in the Notation Section

RRAsh ¼

MInput þ

M Output BA AshFuel þ M Input

MInput

þ Sand

M Output 2Cy þ M Input

AshFuel

M Input þ

Sand

MOutput 1Cy AshFuel þ M Input

M Input

ð1Þ Sand

uðRRAsh Þ RRAsh sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  2  2  2  2ffi uðX Output BA Þ uðX Output 1Cy Þ uðX Output 2Cy Þ uðX Output PM Þ ¼ þ þ þ X Output BA X Output 1Cy X Output 2Cy X Output PM

u0 ðRRAsh Þ ¼

ð2Þ

Each ratio of mass of ash output to the mass of ash in the fuel plus the sand input, XOutput, represents the ash partitioning in the different ash streams (Eqs. (3)–(6)). Sampling uncertainty of PM was considered neglected considering the low budget of PM

X Output

X Output

X Output

¼

BA

1Cy

2Cy

PM

M Input

¼

¼

¼

MOutput BA þ MInput

AshFuel

ð3Þ Sand

MOutput 1Cy þ M Input

Sand

AshFuel

MOutput 2Cy þ M Input

Sand

M Output PM þ M Input

Sand

M Input

AshFuel

M Input

M Input

AshFuel

ð9Þ The recovery ratio, RRi, and partitioning, X Outputi , of each element, i, in each stream was calculated in the same way as the ash recovery ratio, RRAsh RRi ¼

ðM Output BA  C iOutput BA Þ þ ðM Output 1Cy  C iOutput 1Cy Þ þ ðM Output 2Cy  C iOutput 2Cy Þ þ ðM Output PM  C iOutput PM Þ ðM Input AshFuel  C iInput AshFuel Þ þ ðM Input Sand  C iInput Sand Þ

ð10Þ To estimate the uncertainty of each element recovery, a similar calculation to Eq. (2) was used but incorporating the uncertainty of analytical determination of elements according to the law of uncertainty propagation [14–16]. The evaluation of analytical uncertainties was undertaken using data of internal quality control of our laboratory and proficiency tests. The uncertainty for each Ci as estimated according to Eq. (11)

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u2precision þ u2trueness

ð11Þ

Sand

The relative uncertainty, u0 (RRAsh), for this calculation can be estimated with Eq. (2):

X Output

uðX Output PM Þ X Output PM s ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi    2  uMInput AshFuel þMInput Sand uMOutput PM 2 uAshFuelðwÞ 2 ¼ þ þ M Output PM MInput AshFuel þ MInput Sand AshFuelðwÞ

u0 ðX Output PM Þ ¼

uðC i Þ ¼

M Output PM þ M Input

AshFuel

217

ð4Þ

ð5Þ

ð6Þ

Expressions used for relative uncertainties of each ash stream, u0 ðX output Þ, are summarized below uðX Output BA Þ X Output BA s ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi   2     uMInput AshFuel þMInput Sand uMOutput BA 2 uAshFuelðwÞ 2 uBAðwÞ 2 ¼ þ þ þ MOutput BA MInput AshFuel þ MInput Sand AshFuelðwÞ BAðwÞ

u0 ðX Output BA Þ ¼

ð7Þ uðX Output 1Cy Þ X Output 1Cy ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi s   2    ffi uMInput AshFuel þMInput Sand uMOutput 1Cy 2 uAshFuelðwÞ 2 u1CyðwÞ 2 ¼ þ þ þ MOutput 1Cy MInput AshFuel þ MInput Sand AshFuelðwÞ 1CyðwÞ

u0 ðX Output 1Cy Þ ¼

ð8Þ

For the 2nd cyclone the calculation is similar to Eq. (8)

1.3. Evaluation of ash enrichments The volatility of elements present in fuels is usually interpreted as one of the main factors that dictates their partitioning between the different ashes streams produced. Elements with higher volatility are usually found in greater amount in fly ashes and particulate matter collected in the stack. [17]. The enrichment of different elements in ash depends on their chemical association in the fuels and their chemical properties. The determination of enrichment is based on the element ash ratios and the most commons are the relative enrichment, RE, and the enrichment factor, EF. The relative enrichment was initially developed by Meij for coals used in power stations in the Netherlands [17]. It is defined as the ratio between the element i in the chosen ash fraction of the combustor and the concentration of the same element i in the fuel in an ash base.The elements may be grouped in three classes, according to the RE, as presented in Table 1. For biomass the same type of classes may apply. However, because the chemical association of inorganic elements differs considerably from that of coal, the classification of reactivity of elements should be different. An evident example is K which during combustion of certain types of biomass tends to form KCl increasing its volatility. Adjustments of volatility classification may require detailed studies for different types of biomass, as their composition could be substantially variable, particularly what concerns alkalis, chloride and sulphur amounts.RE is a useful tool in predicting tendencies; however, for FBC systems, the EF appears to be more adequate [18,19]. EF is defined as the ratio between the concentration of element i in the chosen fraction of ash (A) to the concentration of the same element i in the fuel, in the ash base, or other ash fraction (B), normalized to the ratio of reference element x present in the two fractions that are being related. EF is given by Eq. (12) and its relative uncertainty, u0 (EF), may be estimated with Eq. (13):

EF ¼

ðC i =C x ÞA ðC i =C x ÞB uðEFÞ EF sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  2  2  2  2 uðC i ÞA uðC x ÞA uðC i ÞB uðC x ÞB ¼ þ þ þ C iA C xA C iB C xB

ð12Þ

u0 ðEFÞ ¼

ð13Þ

218

P. Teixeira et al. / Fuel 101 (2012) 215–227

Table 1 Elements classification for coal based on its behaviour during combustion (adapted from [17]). Class

Behaviour in installation

RE in bottom ashes

RE in fly ashes

Classified elements

I

Elements not vaporized during combustion; their concentration is the same in all ash types

1

1

Al, Ca, Ce, Cs, Eu. Fe, Hf, K, La, Mg, Sc, Sm, Si, Sr, Th, Ti

IIa IIb IIc

Elements that vaporize during combustion and condense within the installation. The RE of bottom ash is less than 0.7 because the elements present in the vapour phase have no chance of condensing on the bottom ash particles. Three subclasses are defined in function the degree of volatility of the element

<0.7 <0.7 <0.7

>4 2< to 4 1.3< to 2

As, Cd, Ge, Mo, Pb, Sb, Tl, Zn Be, Co, Cu, Ni, P, U, V, W Ba, Cr, Mn, Na, Rb

III

Elements very volatile. Generally present in ash to a small degree, and the largest part is present in the vapour phase

<<<1

Compared with RE the use of EF presents two main advantages: (i) it takes into account the dilution effects of sand in ash that escapes from the bed zone to the cyclones and (ii) it corrects possible dilution effects due to different concentration of unburned carbon in the different ash streams. The reference element should not be volatile, which means that its partition between the different ash fractions should be equivalent. It needs to be present in each fraction sampled at different locations and in the case of FBC systems it should be present even in a negligible quantity in the bed material which is usually sand. The selection of a reference element depends generally on the fuel composition, usually being Al, Fe or Ti. As for RE, if EF < 1, which usually happens for some elements in bottom ashes, it means that these elements have been lost from the matrix due to volatilization. If EF > 1, which usually happens in fly ashes, it means that condensation or absorption of volatile elements has occurred in these fractions. If EF  1 elements do not vaporize and their concentration is the same in all ash streams. In previous studies two particular situations were identified as indication of an inadequate selection of a reference element: (i) EF > 1 in bottom ashes may mean that the reference element has been lost from the matrix and the assumption that its concentration remained unchanged in the process is not correct and (ii) EF < 1 in fly ashes and particulate matter it means that the studied element most likely leaves the particles, exiting the installation with the gaseous stream and its condensation in cyclone ashes is unlikely [18]. In the event of agglomeration taking place, an EF > 1 in bottom ashes could mean that there is some agglomeration and the elements have been retained in bed. In this case, it is expected a reduction of the same elements in fly ashes. Analytical uncertainties should be considered for the evaluation of ash enrichment factors. 2. Experimental To evaluate the effect of partial substitution of coal with biomass, the amounts of biomass added were varied from 5% to 25%, on a mass basis. Nevertheless, tests with biomass alone were performed to compare the behaviour of different biomass types during combustion. Three different types of biomass were used: straw pellets (SP), olive cake (OC) and wood pellets (WP). According to CEN/TS 16961 standard, they can be included in different subgroups; herbaceous biomass, fruit biomass and woody biomass, respectively. The use of several biomass types, with different nature, helps on the interpretation of varying behaviours observed during combustion and on establishing correlations with changes in chemical composition. The bituminous coals used were from Europe and South America, namely, Coal Polish (CP) and Coal Colombian (CC). For comparative reasons, tests with coal only were equally performed.

B, Br, C, Cl, F, Hg, I, N, S, Se

2.1. Fuels characterization Physical–chemical properties of the fuels used are presented in Table 2 and 3. The expanded uncertainty (U) associated with each analytical method is also presented, with 95% of confidence, i.e., expansion factor equal 2 (k = 2) [14]. The uncertainty was estimated based on the internal quality control of our laboratory and proficiency tests. Uncertainty is presented for different matrix, biomass and coal, as percentage of the experimental value. 2.2. Description of fuels chemical fractionation methodology The chemical fractionation method used is based on previous studies developed by Zevenhoven [8] and consists of consecutive use of three solvents with increasing solubility capacities, i.e., pure water, 1 M NH4Ac and 1 M HCl. In the first step, to guarantee that the fuel was properly wet, 400 mL of water was used to leach out 20 g of sample (20 mL/g). Polypropylene flask with 500 mL of capacity were immersed in a thermostatic bath, with agitation, at T = 25 °C during 24 h. After this period, the solution was filtered. The remaining fuel collected on the filter was returned to the flask with 400 mL of NH4Ac (20 mL/g), immersed in thermostatic bath with agitation, at T = 25 °C for a period of 24 h. The third and fourth steps were performed with HCl. In the third step the remaining fuel following the filtration was placed in the flask with 300 mL of HCl (15 mL/g) and was immersed in the thermostatic bath, with agitation, at T = 70 °C during 20 h. After this, the suspension was filtered and the remaining solid was put in contact with HCl again (15 mL/ g), for 3 h at T = 70 °C in a ultra-sonic bath. The fourth step was performed because previous studies showed that for some elements one step with HCl was not adequate [8]. The filtration step of suspensions between the different steps was not found to be an easy task for biomass fuels because of the presence of high quantities of colloidal matter. Because of this, filtration was preceded by centrifugation of the suspension. The solid that remains in the filter between steps was washed three times with 20 ml of water, and this water and filtered solution were combined for analysis. To avoid loss of sample between the steps, the acetate cellulose filters used were put together in the containers with the samples. According to our laboratory internal quality control the tests were performed in duplicate and a blank test sample was always carried out. Based on results of repeated tests, instrumental precision and trueness, the uncertainties were estimated. The analytical uncertainties associated with measurements are not illustrated in graphics of Fig. 2. These are referred to relative distribution (%) of elements over the leaching solvents to help the interpretation of results. Differences between the bulk content in fuel (Table 3) and mass distribution over leaching (Fig. 2) can be justified by the analytical uncertainties. The liquid fractions were analyzed by atomic absorption spectrom3 etry for Al, Ca, Mg, Na, K, Fe and Si. The Cl, SO2 4 and PO4 anions were analyzed by capillary electrophoresis for the water soluble

219

P. Teixeira et al. / Fuel 101 (2012) 215–227 Table 2 Proximate analysis, heating values, ash fusibility of fuels and respective uncertainty (95% confidence).

Moisture (a.r., wt.%) Ash (d.b., wt.%)a Volatile matter (d.b., wt.%) Fixed carbon (d.b., wt.%) Low heating value (d.b., MJ/kg) Ash fusibility (oxidant atmosphere) Initial deformation temperature (°C) Softening temperature (°C) Hemispherical temperature (°C) Fluid temperature (°C) a

Standard method

Polish coal

Colombian coal

ASTM D ASTM D ISO 562 ASTM D ASTM D

2.1 6.2 32.2 61.6 28.4

9.3 9.2 37.5 53.3 27.0

1223 1233 1251 1284

1202 1358 1397 1443

3173 3174 3172 5865

ASTM D 1857

Ucoal (%) (K = 2)

Straw pellets

Olive cake

Wood pellets

Ubiomass (%) (K = 2)

5 1 1 6 5

10.6 5.8 76.6 17.6 16.6

7.9 4.9 76.7 18.4 18.9

8.4 0.40 86.2 13.4 18.8

11 4 1 12 6

21 5 5 7

819 1014 1167 1238

751 830 1367 1386

1238 1265 1282 1291

7 3 0.4 3

For biomass samples it was used a temperature of 550 °C (CEN/TS 14774).

Table 3 Elemental analysis of fuels and respective uncertainty (95% confidence) Note: QL-quantification limit. Elemental analysis (d.b., wt.%)

Standard method

Polish coal

Colombian coal

Ucoal (%) (K = 2)

Straw pellets

Olive cake

Wood pellets

Ubiomass (%) (K = 2)

C H N S Cl Al Fe Ca K Na Mg Si P Ti

ASTM D 5373

71.0 4.90 1.2 0.51 0.26 0.69 0.31 0.51 0.10 0.05 0.22 1.01 0.012 0.028

66.2 5.9 1.4 0.65 0.07 1.11 0.45 0.17 0.21 0.011 0.13 2.67 0.006 0.036

1 1 21 3 13 14 12 14 12 15 15 20 18 3

46.7 7.0 0.7 0.14 0.27 0.012 0.010 0.356 1.31 0.029 0.08 1.26 0.080
50.6 7.0 1.1 0.11 0.34 0.011 0.030 0.29 2.11 0.054 0.09 0.06 0.140
49.6 6.9
1 4 6 31 21 39 31 20 24 26 13 19 26 8

ASTM D 4239 ASTM D 2361 ASTM 3682

ASTM 2795 EN 13656

fraction only because the analytical procedure was not adequate for the other leaching solvents. The insoluble fraction was estimated by difference between the total content in the sample and the amount leached out by the three solvents [20]. 2.3. Trial combustion tests The mono-combustion of each fuel and co-combustion tests of binary blends with the two bituminous coals (CC and CP) and three types of biomass (SP, OC and WP) were performed in a pilot fluidized bed combustor for periods of about 3–8 h. The combustor is made of refractory steel and is well insulated with ceramic fibbers. The combustion trial with SP and CP was performed on the old FBC with a square cross-section of 0.09 m2 and 5 m of height. The combustion trials with CC and OC or WP were carried out on a rebuilt FBC with 6 m of height and a cross section area of 0.12 m2. A description of the combustors is given in previous papers [21]. The coals were crushed and sieved and the fraction between 0.25 and 8 mm was used for the combustion tests. The SP and WP were used directly in the form of pellets with 7 mm diameter and maximum of 40 mm length. The OC was already milled and most of the material passed through a 10 mm sieve. For co-combustion runs the two fuels were pre-mixed and the blend was feed with a screw feeder under gravity to the top of the bed. The combustion air was divided between the fluidizing primary air and the secondary air supplied to the freeboard zone. The bed material used was silica sand with 0.37 mm mean diameter. Fresh silica sand was used in each trial and was collected in the end of the each experiment together with the bottom ashes. The schematic representation of pilot fluidized bed is given in Fig. 1.

Fig. 1. Scheme of the new fluidized bed combustor of LNEG.

P. Teixeira et al. / Fuel 101 (2012) 215–227

HCl

NH4Ac

Residual

H2O

100% 80% 60% 40% 20% 0% SP OC WP SP OC WP SP OC WP SP OC WP SP OC WP SP OC WP SP OC WP

Al

Ca

Mg

Residual

Na

HCl

K

Fe

NH4Ac

20% 0% Al

CP

CC

Ca

CP

CC

CP

Mg

CC

Na

CP

CC

K

20000 15000 10000 5000 0 SP OC WP SP OC WP SP OC WP SP OC WP SP OC WP SP OC WP SP OC WP

CP

CC

CP

Fe

Ca

Mg

Residual

40%

CC

H2O

25000

H2O

60%

NH4Ac

30000

Al

80%

CC

HCl

35000

Si

100%

CP

Mass Distribution over leacging solvent (mg/kg)

Residual

Mass Distribution over leaching solvent (mg/kg)

Distribution over leaching solvents

Distribution over leaching solvents

220

Na HCl

K

Fe

NH4Ac

Si H2O

35000 30000 25000 20000 15000 10000 5000 0 CP

CC

Al

Si

CP

CC

Ca

CP

CC

CP

Mg

CC

Na

CP

CC

CP

K

CC

Fe

CP

CC

Si

Fig. 2. Relative distribution (%) and mass distribution (mg/kg) of elements by chemical fractionation.

Table 4 presents the operating conditions used for the three set of combustion runs that will be called SP/CP, OC/CC and WP/CC. The ashes were collected in four locations as defined in Section 1.2; BA that includes ash and bed sand material, 1Cy, 2Cy, and PM that was sampled with quartz fibre filters of an isokinetic sampling train (method US EPA 29) inserted in the stack [22] during about 1 h. Ash characteristics are presented in Table 5. Particle size diameters (PSD) of BA are mainly dictated by the sand used and also by the characteristics of bed ashes formed. PSD of cyclone ashes was measured with Malvern laser diffraction and was found to be larger for particles collected in the 1st cyclone than those collected in the 2nd cyclone. PM escaped to the stack was sized with a Mark III Cascade impactor which revealed PSD smaller than 10 lm. In FBC systems the characteristics of each particle stream are influenced by the sand used, duration of combustion tests, mass flow and type of fuel used. Amounts of bed sand used in the SP/CP runs were about 20 kg and for the other runs in the new installation higher amounts were used, 100 kg, which produce different sand to ash ratios. In the case of SP/CP tests higher loads of ash in the fuel input to sand (0.15–0.33 kg ash/kg sand) were obtained than for OC/CC tests (0.02–0.04 kg ash/kg sand). For WP/CC these ratios

Table 4 Combustion conditions of the pilot fluidized bed.

Table 5 Characteristics of ash streams from LNEG pilot fluidized bed. Identification

Particle size diameter

BA

Collected at the boiler bottom

Coarse

1Cy

Removed from flue gases with a first cyclone

Between 5.8 and 564 lm Depending of fuel the d50 varies between 28.5 and 93.9 lm

2Cy

Removed from flue gases with a second cyclone

Between 1.2 and 118 lm Depending of fuel the d50 varies between 7.5 and 12.0 lm

PM

Removed from the flue gases with quartz fiber filters on an isokinetic sampling train

Between 0 and 12.4 lm Depending of fuel the d50 varies between 0.7 and 3.4 lm

were even lower, 0.002–0.04 kg ash/kg sand, due to the very low ash content of wood. BA is mostly affected by these ratios although fly ash, especially those collected in the 1st cyclone, may also be affected by the ratios of fuel ash input to sand. 2.4. Mass balances and elements partitioning in ash streams

Fuel blend

SP/CP

OC/CC

WP/CC

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

9.6–15.7 269–233 701–818 0.7–1.1 1.1–1.3 31–52 18–35 0.16–0.18

11.5–20.3 280–253 766–850 0.7–1.0 0.9–1.5 35–45 23–25 0.5

11.5–18.3 280–316 831–850 0.7–0.8 0.9–1.2 36–38 23–33 0.5

Four balances, with different resolution and capacity, were used for weighing steps. The uncertainty of measurements was estimated based on the repeatability and resolution of the balances used. The uncertainty associated with the balance calibration was considered negligible and systematic errors identified were incorporated in the uncertainty estimation. The evaluation of uncertainties of the analytical determination of ash in the fuel and loss on ignition were carried out using data of internal quality

221

P. Teixeira et al. / Fuel 101 (2012) 215–227

control and proficiency testing. The same approach was applied to estimate the uncertainty of the analytical determination of elements. Concentration of elements in ash streams was performed in accordance with standard procedures. For Al, Ca, Mg, Na, K, Fe and Si analysis, ash samples were fused at 1000 °C with lithium methaborate and dissolved with diluted HCl in accordance with ASTM 3682. Elements were quantified with Atomic Absorption Spectrometry. Levels of C and S were measured to account for the levels of CO2 and SO3 in ash and verification of completeness of ash composition determination. Cyclone ashes were previously calcined at 750 °C to release unburned carbon that may interfere with the fusion process. For PM the elements were measured in the digested filters of the isokinetic stack catching (EPA no. 29). The ash mass recoveries ratios in the combustion tests, RRAsh, and relative uncertainties, u0 (RRAsh), were estimated with Eqs. (1) and (2). The partitioning, XOutput, of the four ashes streams were estimated with Eqs. (3)–(6) and the relative uncertainties, u0 (Xoutput), were estimated with Eqs. (7)–(9). The results are presented in Table 6. The recovery ratio of each element, RRi, was estimated in accordance with Eq. (10) and the results are given in Fig. 5. Mass balances and elements partitioning were not performed for Cl, S and P, because, for some ash streams there are analytical limitations (values lower than QL) and also because a significant quantity of S and Cl are present in the gaseous phase that was not considered in this study dedicated to ash.

3. Results and discussion 3.1. Fuels chemical fractionation The results of the chemical fractionation of Al, Ca, Mg, Na, K, Fe and Si in biomass and coals (Fig. 2) showed significant differences regarding the distribution of elements with leaching solvents. The use of two HCl extractions steps improved the total HCl extraction between 2% and 50%. The second extraction was more relevant for coals than for biomass, especially for Fe, K, Al and Si. It was observed that for CP and CC the second step of HCl allowed extracting respectively, more 12% and 8% of Fe, 3% and 1% of K, and 2% of Al and 1% of Si for both coals. In the case of biomass the second step was not so relevant; nevertheless for SP it allowed extracting more

8% of Fe and 4% of Al. However, given the low concentrations of certain elements or low HCl solubility these quantities may be considered negligible in most of the cases, hence no great benefit arose from this additional step besides ensuring the extraction efficiency. The comparison of coal and biomass, in Fig. 2, reveal that almost every element appears to be more easily leached out by water and NH4Ac from biomass than in coals tested. This means that all the species of biomass used have a significant proportion of the ash forming elements (probably in ionic forms or associated with organic phases) that tend to easily flake off from the matrix during combustion. This was especially evident for the alkalines, K and Na, and also for alkaline-earth elements, Ca and Mg, which were almost completely extracted in the two first fractionating steps. However, as observed in Fig. 2, only the soluble phases of K and Ca may have a dominating role, because of the quantities involved, this being especially the case with K. Regarding the other three elements, Al, Fe and Si, different patterns can be observed for each species of biomass, although in amounts, only Si appears in significant quantity in SP, which is due to its nature. In the case of OC Al, Fe and Si were partially extractable with water, about 30% of Al and Fe and only 15% of Si. Availability of Al in soils and its uptake by plants is especially dependent on pH conditions, increasing with a decrease of soil pH [23]. The OC used in the combustion trial came from Baixo Alentejo in Portugal that present soil pH values moderately acidic (5.6–6.5) or very acidic (4.6–5.5), which may explain the Al behaviour. Fe soluble in water could arise from the presence of iron sulphate or soluble chelates. This was more evident for OC, which eventually can be related with the fertilization programs sometimes used in the olive plantations to avoid the chlorosis of leaves, the small and yellowish fruits induced by iron deficiency [24]. In SP and WP more than 90% of these elements appear as acid soluble or residual resembling more refractory species. In coals, as shown in Fig. 2, all the Al, Fe and Si were only extractable in the HCl step or remained in the residual fraction. This suggests the presence of aluminium silicates, such as clays, that also justify the K bound to the residual fraction or slightly solubilized by HCl. Presence of iron sulphides may justify the major acid soluble phase of Fe. Ca and Mg resemble similar solubility in each coal, being about 15% of Ca and Mg soluble in water and NH4Ac in CP and 60% and 51% of Ca and Mg respectively soluble in the CC. Although in very small amounts in coal, about 59% and 87% of the Na was extractable with water and NH4Ac in CP and

Table 6 Ash partitioning, XOutput and ash mass recovery ratio, RRAsh, of FBC combustion test. Straw pellets/Coal Polish combustion runs XOutput_BA (%) XOutput_1Cy (%) XOutput_2Cy (%) XOutput_PM (%) RRAsh (%)

100% CP 90 ± 3 8.9 ± 0.4 1.37 ± 0.04 0.42 ± 0.01 101 ± 4

5% SP 97 ± 4 8.9 ± 0.4 1.41 ± 0.06 0.31 ± 0.01 107 ± 9

15% SP 86 ± 4 7.9 ± 0.4 1.16 ± 0.05 0.93 ± 0.04 96 ± 8

25% SP 103 ± 4 2.8 ± 0.1 0.79 ± 0.03 0.49 ± 0.02 107 ± 9

100% SP 74 ± 3 2.9 ± 0.1 0.25 ± 0.01 1.42 ± 0.06 79 ± 7

100% CC 97 ± 2 2.0 ± 0.1 0.35 ± 0.01 0.149 ± 0.003 99 ± 3

5% OC 98 ± 4 2.6 ± 0.1 0.32 ± 0.01 0.063 ± 0.003 101 ± 9

15% OC 98 ± 4 3.1 ± 0.1 0.36 ± 0.02 0.076 ± 0.003 102 ± 9

25% OC 95 ± 4 2.7 ± 0.1 0.31 ± 0.01 0.056 ± 0.002 98 ± 8

100% OC 99 ± 4 3.0 ± 0.1 0.19 ± 0.01 0.28 ± 0.01 103 ± 9

5% WP 94 ± 4 2.3 ± 0.1 0.35 ± 0.02 0.043 ± 0.002 97 ± 8

15% WP 97 ± 4 2.6 ± 0.1 0.42 ± 0.02 0.107 ± 0.002 100 ± 9

25% WP 100 ± 2 2.4 ± 0.1 0.35 ± 0.02 0.123 ± 0.003 102 ± 9

100% WP 98 ± 4 1.7 ± 0.1 0.09 ± 0.01 0.0105 ± 0.0002 100 ± 9

Olive cake/Coal Colombian combustion runs XOutput_BA (%) XOutput_1Cy (%) XOutput_2Cy (%) XOutput_PM (%) RRAsh (%) Wood pellets/Coal Colombian combustion runs XOutput_BA (%) XOutput_1Cy (%) XOutput_2Cy (%) XOutput_PM (%) RRAsh (%)

222

P. Teixeira et al. / Fuel 101 (2012) 215–227

CC respectively, which means that if Na was available in greater amounts it could present ash problems during combustion. The solubility of Cl, S and P in water at 25 °C was also evaluated. Fig. 3 presents the solubility in water as % of the bulk contents. Solubility of Cl and S was always inferior for coal than for biomass. For coals, P was present in low amounts and was not found in soluble forms. For the biomass species studied, soluble P varied between 50% and 76%, which meant that a significant quantity of P was present as soluble phosphates. Absolute quantities are, however, of concern in OC and SP, but not in WP that present very low concentration of P. Levels of S on the bulk contents of both coals were found to be similar, although in CC a significant solubility (24%) indicates the presence of different compounds, perhaps sulphates. S is usually present in coals as sulphides and pyrites with very low solubility or organically associated, which mean that beside the low solubility the tendency to volatilize and form sulphur oxide or sulphates usually depends mostly on interactions with alkaline and earth alkaline metals. For biomass, S was only detected in SP and OC, 0.14% and 0.11% respectively, of which 43% and 23% were found to be soluble in water. Nevertheless, as for coals, it is expected that the release of S to the gaseous phase or formation of sulphates depends on the composition of ashes and thermodynamic conditions of boilers. The coals present different amounts of Cl, being considerably high in CP as for two of the species of biomass used. About 20% was found to be soluble in water for both coals which means that Cl is mostly insoluble or might be present in organic association. For the two biomass species, SP and OC, all the Cl content was found to be soluble in water, which means that Cl was present only as ionic salts. In the case of SP the water soluble chloride was found to be higher than the bulk chloride content (obtained by bomb combustion method). This fact could not be justified only by the analytical uncertainties associated with the quantification methodology. Differences between water soluble Cl (120 °C) and total Cl have already been found in the BioNorm Project [25] and were further studied in BioNorm II [26,27]. It was shown that major differences between Cl obtained by bomb combustion method and water soluble content occur for biomass high in ash and chlorine. This may be due to an undesirable effect during the bomb combustion method that lead to retention of Cl compounds in the residual ash matrix produced and hence is not absorbed in the collecting solution used in the bomb that is subsequently analyzed for Cl. 3.2. Ash mass balance and partitioning Ash mass balances were performed for each combustion tests, considering the ash mass inlet as a sum of ash in the fuel mass plus

Cl

S

CC

SP

P

Soluble Fraction in Water

200%

150%

100%

50%

0% CP

OC

WP

Fig. 3. Water solubility (25 °C) fractions of Cl, S and P in coals and biomass.

the sand used, and the ash mass streams output. Mass of cyclone ashes were considered on a LOI free base and for PM, total masses were calculated based on the entire flue gas volume produced during the test run and the PM emissions measured without any flue gas treatment besides cyclone dedusting. Ash partitioning and mass recoveries ratios are expressed in % in Table 6. The uncertainty associated with ash mass balances (Table 6) has also a direct bearing on difficulties related with the quantification of total mass of various elements. Hence, possible lack of recovery ratios of the various elements should be evaluated taking into account uncertainties associated with mass balances. The recovery of ash varied between 96% and 107%, except in the case of mono-combustion of SP. Assuming the uncertainty (95% confidence) associated with the ash mass balances, between 3% and 9%, it can be concluded that all mass balances can be closed, except in case of the 100% SP combustion. In this case, the lack of closure seems to be related with the agglomeration effects of bottom ashes that turned difficult to remove some agglomerates from the bed of the FBC. Though this uncertainty approach does not take into consideration possible variations due to operation parameters, it may provide useful information about ash mass balance variability. In fact, it was interesting to verify that the uncertainty associated with the mass balance of mono-combustion of biomass or biomass blends was higher than when only coal was used, which may be explained by the higher uncertainty involved in the biomass ash content determination (Table 2). Results of the ash partitioning at four locations revealed that most of the mass was retained in the bottom of the bed which could be explained by the large amount of sand used as bed material compared with the total mass of ashes produced. In the old FBC installation, where SP trial was performed, the % of ashes in the 1st cyclone varied between 3% and 9% and on the new installation this decreased to 2–3% of the total ash. This may be explained by higher height of the new installation that may favour higher retention of sand and ashes inside the combustor decreasing elutriation. Generally, the % of ashes retained in the 2nd cyclone during coal monocombustion and co-combustion varied between 11% and 18% of ashes retained in the 1st cyclone (except for co-firing 25% SP). When biomass is combusted alone, the percentage of ashes retained in the 2nd cyclone varied between 5% and 11% of ashes retained in the 1st cyclone, which can be related with the size of ash particles. During biomass mono-combustion the ash particles formed are smaller than the particles formed during coal monocombustion and co-combustion, more PM escaped from the 2nd cyclone to the stack. An increase of mass of PM was observed for SP and OC mono-combustion. 3.3. Combustion tests performance based on LOI A major characteristic of bubbling FBC without recirculation of elutriated particles is the presence of unburned material in fly ash due to incomplete combustion. Fig. 4 illustrates the variation of LOI (loss on ignition, performed at 750 °C) of the different ashes collected in cyclones. Ash from coal mono-combustion presents the largest LOI contents, but some ash from co-combustion tests also has exhibited considerable LOI. Although the gas velocity in the freeboard was inferior to the terminal velocity of the particles estimated for the LNEG pilot FBC, which is 2.1 m/s [28], the LOI (that was similar to the carbon content) of the cyclone ashes varied between 20% and 60% in the case of coal mono-combustion and cocombustion. For biomass mono-combustion the carbon content and LOI were usually lower than 7%. The low LOI levels in biomass ash streams are coherent with the high volatile content of biomass that results in mostly gaseous phase and fast combustion. For OC/ CC and WP/CC co-combustion runs the LOI was found to be always higher in the 2nd cyclone. This can be related with the low density

P. Teixeira et al. / Fuel 101 (2012) 215–227

Loss on Ignition

60%

40%

20%

1Cy 100 CP 2Cy 100 CP 1Cy 5 SP 2Cy 5 SP 1Cy 15 SP 2Cy 15 SP 1Cy 25 SP 2Cy 25 SP 1Cy 100 SP 2Cy 100 SP 1Cy 100 CC 2Cy 100 CC 1Cy 5 OC 2Cy 5 OC 1Cy 15 OC 2Cy 15 OC 1Cy 25 OC 2Cy 25 OC 1Cy 100 OC 2Cy 100 OC 1Cy 100 CC 2Cy 100 CC 1Cy 5 WP 2Cy 5 WP 1Cy 15 WP 2Cy 15 WP 1Cy 25 WP 2Cy 25 WP 1Cy 100 WP 2Cy 100 WP

0%

Fig. 4. LOI (%) in ashes of the 1st and 2nd cyclones.

of the small carbonaceous particles, which could have formed during the devolatilization phase, that are quickly carried away from the hot FBC freeboard and subsequently cool down before thought complete combustion. SP/CP ashes, except for 100% SP run, were observed to have larger LOI levels which may be due to the smaller height of the old FBC installation and use of smaller sand quantities. In the old installation, the fuel time residence was shorter and subsequent influence was more pronounced for coals. Although quantification of elements for mass balance was performed in LOI free ashes, the influence of ash elutriation could have significant effect on the behaviour of ash forming elements. The presence of unburned and very porous carbon gave rise to low density originating particles which also contain the mineral material. Hence, minerals that could be expected to remain as bottom ash were found as light fly ash particles due the elutriation. For example, Al in SP and CP was extracted only by HCl, hence it may be considered a non-reactive element; however, Al was found in the cyclone ash. The same happened for Fe in coals, but its presence in cyclone ash was also significant. 3.4. Elements balance and partitioning in ash streams Mass balances were carried out for seven elements, and the % recovery ratios and their uncertainty levels are presented in Fig. 5. As can be seen in the figure uncertainties are largely responsible for the apparent lack of closure in mass balances for most elements. For SP mono-combustion, even taking into account uncertainties, the recovery ratios of Ca, Mg, and K were poor which can be associated with agglomeration as was verified. The Al and Fe recovery ratio for 25% SP was lower than 100%. A possible explanation can be related with heterogeneity of feeding during co-combustion runs. If the proportion of SP introduced was a slightly higher than 25%, expected variations in concentration of elements mentioned before occurred, especially when the differences in the amounts of fuel was large as was the case of Al and Fe in SP and CP. This effect was not detected in ash mass balance because the ash contents in CP and SP are similar. For OC/CC tests the uncertainties allowed the closure of all elemental mass balance, which means that apparently the input and output of elements in FBC was well controlled, as well as the chemical analysis. Nevertheless, some recovery ratios were found high (e.g. Ca in 100% CC and 5% OC, and Fe in 100% OC) and this requires a detailed analysis regarding partitioning of elements in different ash streams. In the case of WP/CC combustion runs, mass balance suggested that beside Na, there are also signs of contamination for Ca and K, being in generally worst for WP mono-combustion. A possible reason for this could be the low ash content of WP and the different

223

concentration of these elements in the several fuels processed in the installation. Any cross contamination of the ash streams from previous test runs can be easily detected, especially when using woody fuels with low mineral content. This type of problem was already identified in literature, where ash mass recovery reached 160% [10]. Nevertheless, for most of the combustion tests this contamination was not observed, and when took place appeared to be correlate with the more reactive elements, Na, K and Ca, in accordance with the chemical fractionation (Fig. 2). Given the acceptable mass balances generally attained, the partitioning of elements in ash streams was established and is presented in Fig. 6 as % for each element recovery. From Fig. 6 it can be concluded that the amount of elements in the PM fraction is not significant; except for Na and K for 25% SP and 100% SP runs. The differences between the partitioning of elements in bottom ashes of the old installation combustor and the new one are equally evident, it was verified that a higher proportion of elements are retained in bottom ashes of the more recent installation. During the replacement of CP by SP, the fraction of elements making up ash, found in BA increase and consequently the fraction of the same elements in cyclones decrease. This may be due, in part, with higher combustion efficiency and thus less elutriation of unburned fuel particles. SP contains equivalent Si, much larger quantities of K but less amounts of the other elements compared to ash of CP. Although K was observed to be essentially reactive it was not found solely in the cyclones and PM ashes. This suggests that it was not completely volatilized, which may be explained by reactions involving agglomeration in the bed [29], as it was detected. A major observation is the presence of Na and K in the PM fraction for 25% SP and 100% SP. This can be explained by the reactivity of Na and K in SP as observed in Fig. 2, which may form chlorides and sulphates given the presence of reactive S and Cl (Fig. 3). Fe was not identified as a volatile element according to chemical fractionation of SP and CP; hence its presence in PM could probably be due to possible contaminations from the fluidized bed reactor surfaces. When replacing CC with OC levels of K increased significantly due to its higher content in OC. Although K was found to be extremely reactive its presence in cyclones and PM was lower than expected as most of it was retained in BA, especially in the 100% OC. In the OC mono-combustion K retention in BA increased 40% comparing with the 25% OC run, which may be related with K reaction with the sand bed [29]. Ca partitioning for 100% CC and 5% OC appear to be in discrepancy probably because of difficulty in closing elemental mass balances, in spite of values are between the uncertainty limits. However they are significantly higher than 100%, which can be justified by Ca retention in BA in greater amounts in these combustion tests than in 15% OC and 25% OC. The partitioning of Fe in the OC/CC runs was presented a different behaviour, i.e., its retention in BA did not grow so substantially with the increase of OC as the other elements. This can be related with the reactive Fe that was found in chemical fractionation for OC (about 30%) or with possible contamination as referred for SP/CP. For the WP/CC runs the strong differences between ash contents and concentration of elements of WP and CC was not found to result in significant differences in the partitioning of elements when the share of WP increased, as can be observed in Fig. 6. It was verified that the introduction of biomass up to 25% did not influence in a significant way the partitioning of elements in ashes because most of elements originated from the coal. Only with 100% WP it was verified an alteration in the partitioning of elements in ashes from different locations. It was observed an increase of retention in BA, which can be attributed to a more efficient combustion of biomass compared to coal and thus less elutriation of particles.

224

P. Teixeira et al. / Fuel 101 (2012) 215–227

reference element were performed. Results are presented in Fig. 7 with estimated uncertainties levels. However, Al introduces a certain degree of inaccuracy for some streams, especially for biomass mono-combustion due their lower content in Al. Another problem of using Al is related with its significant presence in the sand used, 0.34%, hence the Al of the sand had to be considered in the ash enrichment factor, i.e., there was an addition to Eqs. (12) and (13) corresponding to the Al content of sand. It was also assumed that Al in the fuels was not volatile, although for OC it was detected some reactivity. Only the BA and cyclone ash enrichment factors were evaluated because for most of the cases there was insufficient data for PM. As can be observed in Fig. 7 the main observations are: Ca and Mg enrichment in the 2nd cyclone ashes for 100% OC and WP, which can be related with their reactivity as observed in chemical fractionation analysis and possible formation of sulphates or phosphates. The presence of P is especially relevant in the case of OC not only because of its significant presence but also

Similar to OC/CC runs, a higher retention of Ca and Na was observed in BA (50%) for all WP/CC combustion tests. It may partly be explained by difficulties in closing mass balance. Another possible reason could be the reactivity of the two elements in both fuels as it was evidenced by the chemical fractionation. The reactive fraction of Ca and Na may suggest that those elements may react in the bed leading to the formation of silicates. The formation of calcium silicates is referred to by several authors [30,31] which imply retention of Ca in bed. Formation of ternary eutectics with low melting point, formed by K2O–Na2O–SiO2 (about 540 °C), is also mentioned in the literature [31]. 3.5. Ash enrichment Relative enrichments of elements at four locations of FBC were calculated; however, results need to be analyzed with caution because of dilution due to sand elutriation. Hence, the calculation of the ash enrichment factors for Ca, Mg, Na, K and Fe using Al as a

Recovery ratio

300% Straw Pellets Coal Polish

200%

100%

0%

Al

Ca

Mg

Na

K

Fe

Si

300%

Recovery Ratio

Olive Cake Coal Colombian

200%

100%

0%

Al

Ca

Mg

Na

K

Fe

Si

300%

Recovery Ratio

Wood Pellets Coal Colombian

200%

100%

6

0%

Al

Ca

Mg

Na

K

Fe

Fig. 5. Recovery ratio of elements, RRi, in the combustion tests.

Si

225

P. Teixeira et al. / Fuel 101 (2012) 215–227

BA

1Cyc

2Cyc

PM

Partitioning in Ash

100% 80% 60% 40% 20% 0%

Al

Ca

Mg

BA

1Cyc

Na

2Cyc

K

Fe

PM

Partitioning in Ash

100% 80% 60% 40% 20% 0%

Al

Ca

Mg

BA

1Cyc

Na

2Cyc

K

Fe

K

Fe

PM

Partitioning in Ash

100% 80% 60% 40% 20% 0%

Al

Ca

Mg

Na

Fig. 6. Partitioning of elements in ash streams of the SP/CP, OC/CC and WP/CC combustions runs.

because 75% of it is water soluble. In the case of WP about 50% of P was also water soluble (S and Cl were lower than QL). Enrichment of Ca in bottom ashes and its depletion in the 1st or 2nd cyclone was observed for 100% CC and 5% OC, which clearly demonstrates differences between those fuels. Although part of Ca is reactive in CC it was mostly retained in BA, probably as CaSO4 due to the significant S content in coal. In general, a tendency for Na enrichment in bottom ashes was observed, especially for 100% CC, 5% WP and 100% WP combustion tests, and there was no enrichment of Na in cyclone ashes. This retention in bottom ashes may be possible related with the formation of low melting point eutectics such as K2O–Na2O–SiO2, as mentioned before [31]. Na and K depletion or low enrichment in all ash streams for 100% SP combustion was observed which could be associated with the lack of closure of ash balance. These elements are often constituents of bed agglomerates that are difficult to recuperate

quantitatively. It was showed by SEM/EDS analysis of the agglomerates from LNEG SP combustion tests, that K was one of the main elements of the agglomerates [29]. K enrichment in the 1st cyclone and 2nd cyclone in 100% WP was observed, which may have to do with the high volatility of K and absence of an ash matrix that can retain K in BA. The releasing of inorganic elements to the gaseous phase during combustion of woody fuels, in spite of its lower amounts compared to shorter rotation biomass was already mentioned in the literature [32]. Depletion of K in the 1st and 2nd cyclone ashes in the case of 100% OC was observed, although no significant enrichment was observed in BA. Zevenhoven [33] reported that about 60% of the reactive K may react with the bed material when sand is used and wood derived fuels are fired, which means that K available for deposit formation in cooler convective transfer zones could be overestimated. The partitioning of reactive K between bottom ashes and cyclones is strongly dependent on fuel bulk characteristics and bed material

Sodium

6 5 4 3 2 1 0

1Cyc

7 6 5 4 3 2 1 0

BA

1Cyc

7 6 5 4 3 2 1 0

BA

1Cyc

2Cyc

1Cyc

Sodium

3 2 1 0

BA

1Cyc

7 6 5 4 3 2 1 0

BA

1Cyc

7 6 5 4 3 2 1 0

1Cyc

2Cyc

Enrichement Factor Enrichement Factor

1Cyc

2Cyc

Magnesium

6

BA

1Cyc

7 6

2Cyc

Sodium

5 4 3 2 1 0

BA

1Cyc

7 6 5 4 3 2 1 0

2Cyc

Potassium

BA

2Cyc

Iron

BA

7 6 5 4 3 2 1 0

2Cyc

Potassium

Calcium

BA

2Cyc

5 4

2Cyc Iron

6

7 6

2Cyc

Potassium

Magnesium

BA

2Cyc

7 6 5 4 3 2 1 0

2Cyc

Enrichement Factor

Enrichement Factor

7

BA

Enrichement Factor

1Cyc

7 6 5 4 3 2 1 0

1Cyc

Enrichement Factor

6

Calcium

BA

Enrichement Factor

Magnesium

BA

Enrichement Factor

2Cyc

Enrichement Factor

7 6 5 4 3 2 1 0

1Cyc

Enrichement Factor

Enrichement Factor

BA

7 6 5 4 3 2 1 0

Enrichement Factor

Calcium

Enrichement Factor

7 6 5 4 3 2 1 0

Enrichement Factor

P. Teixeira et al. / Fuel 101 (2012) 215–227

Enrichement Factor

226

1Cyc

7 6 5 4 3 2 1 0

2Cyc

Iron

BA

1Cyc

2Cyc

Fig. 7. Ash enrichment factors of elements in ash streams of the SP/CP, OC/CC and WP/CC combustion runs.

used. This fact may explain that in spite of K being almost completely soluble in water an effective enrichment of it was not observed in any ash stream for OC combustion test. Usually is the availability of Si from the fuel, as in the case of SP, that is related with a high retention of K in BA as potassium silicates [34,35]. However K may also react with Si from sand. Nevertheless this process seems to be slower compared with reactions with Si from herbaceous biomass, as it was suggested by the different behaviour of bed for SP and OC combustion. It was observed a quicker bed defluidization for SP mono-combustion compared with OC, although for

OC mono-combustion some enlargement of sand particles and weak sand bridges were also visible. It is possible that slower reactions between K and sand may limit K fixation in BA. Fe enrichment was observed in cyclone ashes when only biomass was used because of possible contamination. 4. Conclusions The ash behaviour of five different fuels under monocombustion and co-combustion conditions were evaluated using

P. Teixeira et al. / Fuel 101 (2012) 215–227

three approaches: chemical fractionation, partitioning of elements making up ash and enrichment of these elements in several ash streams. Chemical fractionation may give an indication of reactivity of elements present in fuels. Usually a high reactivity means that those elements can rapidly volatilize and then may create fouling problems. The role of elements like Cl that may promote the release of elements to the gaseous phase and subsequent formation of deposits is already known. Depending on the bulk composition of the ashes the reactivity/volatility of the elements is also related with the tendency to form silicates with low melting points in the bottom ash. This situation is particularly important in FBC systems, in which apart from bottom ash, sand may also interfere in reactions influencing the partitioning of elements. However, other factors may have a strong influence, as for example the fuel elutriation during combustion, which can upset mass partitioning. This effect was more pronounced during SP/CP combustion tests performed on an FBC shorter in height. The design and operation mode (e.g. fluidizing agent amount) of FBC systems also interfere with the behaviour of elements. Because of the complexity of FBC systems the chemical fractionation could not determine entirely the behaviour of the elements during combustion. The ash enrichment factor was found to be useful in identifying the affinity of elements to different ash streams and helped to understand possible reaction between different elements. The main observations was the Ca and Mg enrichment in cyclones when 100% OC and 100% WP were burned alone (maybe as phosphates or sulphates), and Ca retention in bottom ash for 100% CC and 5% OC (maybe as calcium silicates). K behaviour was found to be more variable for each type of biomass studied. The fruit biomass (OC) presented a high quantity of reactive K and small amount of Si, which means that retention in the bottom ash, was not very pronounced because the reaction between K and Si from sand appears to be slower than reactions between the K and Si coming from ash fuel matrix, that occurs in some types of herbaceous biomass like straw. In the woody biomass (WP), given the small amount of ash, the capacity to retain K in bottom ash appeared to be weak; hence, more K was found in cyclones. For the herbaceous biomass (SP) enrichment was expected in bottom ash. Nevertheless, due to agglomeration problems the lack of closure of ash mass balance limits clear conclusions. Difficulties in closing elemental mass balance may be explained by the segregation of K in these agglomerates as mentioned before. For a better understanding of the behaviour of elements during combustion, thermodynamic studies to predict the formation of compounds, based on chemical fractionation are being carried out for the experimental results presented in this paper. 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 equally acknowledged. References [1] Directive 2009/28/EC of the European parliament and of the council of 23 April 2009 on the promotion of the use of energy from renewable sources. [2] Baxter L. Biomass-coal co-combustion: opportunity for affordable renewable energy. Fuel 2005;84:1295–302.

227

[3] CEN/TS 14588. Solid biofuels – terminology, definitions and descriptions; 2003. [4] CEN/TS 14961. Solid biofuels – fuel specifications and classes; 2005. [5] Benson S, Holm P. Comparison of inorganic constituents in three low-rank coals. Ind Eng Chem Prod Res Dev 1985;24:145–9. [6] Miller T, Baxter L. Alkali deposits – found in biomass power plants – a preliminary investigation of their extent and nature, vol. I. US Department of Energy; 1996. [7] Miller T, Baxter L. Alkali deposits – found in biomass power plants – a preliminary investigation of their extent and nature, Vol. II. US Department of Energy; 1996. [8] Zevenhoven M, Skrifvars B, Yrjas P, Hupa M, Nuutinen L, Laitinen R. Searching for improved characterization of ash forming matter in biomass. In: 16th International conference on fluidized bed combustion; 2001. [9] Puttkamer T. Charakterisierung biogener Festbrennstoffe. Stuttgart University; 2005. [10] 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: Co combustion with wood. Energy Fuels 2010;24:333–45. [11] Zevenhoven-Onderwater M. Forming matter in biomass fuels. Academic Dissertation, Åbo Akademi University; 2001. [12] Kumar V. Investigation into ash related issues during co-combustion of coal and biomass: development of a co-firing advisory tool. Doctor of Philosophy Thesis, Curtin University of Technology; 2007. [13] Ward C, French D. Analysis and significance of mineral matter in coal. In: 21st Annual meeting of the society for organic petrology; 2004. [14] Evaluation of measurement data – guide to the expression of uncertainty in measurement. JCGM 100:2008. [15] Eurachem/CITAC Guide. Quantifying uncertainty in analytical measurement, 2nd ed., 2000. [16] Nordtest Tr 537. Handbook for calculation of measurement uncertainty in environment laboratories, 2nd ed., 2004. [17] Meij R, Winkel H. The emissions of heavy metals and persistent organic pollutants from modern coal-fired power stations. Atmos Environ 2007;41:9262–72. [18] Sorbo N, Tchobanoglous G, Lucero J. Fate of selected metals and emissions from a sludge/waste paper gasifier. EPA Project summary; 1986. [19] Marani D, Braguglia C, et al. Behaviour of Cd, Cr, Mn, Ni, Pb, and Zn in sewage sludge incineration by fluidised bed furnace. Waste Manage 2003;23:117–24. [20] Zevenhoven M, Yrjas P, Backman R, Skrifvars B, Hupa M. The abo akademi database – fuel characterization. In: 18th International conference on fluidized bed combustion; 2005. [21] Lopes H, Gulyurtlu I, Abelha P, Crujeira T, Salema D, Freire M, et al. Particulate and PCDD/F emissions from coal co-firing with solid biofuels in a bubbling fluidized bed reactor. Fuel 2009;88(12):2327–532. [22] Method EPA 29. Determination of metals emissions from stationary sources; 1996. [23] Ruszczyn´ska A, Bien´kowski P, Bulska E. Fractionation of total water soluble aluminum in extracts from peat and sand soil samples by ion-exchange method followed by GF-AAS. Int J Environ Studies 2005;62:193–200. [24] Olives Plantation Fertigation Guide, [acceded June 2010]. [25] BioNorm – Pre-normative work on sampling and testing of solid biofuels for the development of quality assurance systems. BioNorm project – final technical report; 2005. [26] BioNorm II – Pre-normative research on solid biofuels for improved European standards: improved and extended best practice guideline for the determination of Sulphur and halogens; 2010. [27] BioNorm II – Pre-normative research on solid biofuels for improved European standards: best practice guideline for the use of XRF for solid biofuels; 2010. [28] Abelha P. Emissões gasosas durante a co-combustão de carvão com resíduos. Universidade Nova de Lisboa. Tese doutoramento em Engenharia Química; 2004. [29] Teixeira P, Lopes H, Gulyurtlu I, Lapa N. Prediction of Slagging and fouling tendency of biomass co-firing in fluidized bed combustion; Unpublished Results. [30] Thy P, Lesher C, Jenkins B. Experimental determination of high-temperature elemental losses from biomass slag. Fuel 2000;79:693–700. [31] Risnes H, Fjellerup J, Henriksen U, Moilanen A, Norby P, Papadakis K, et al. Calcium addition in straw gasification. Fuel 2003;82:641–51. [32] Lith S, Jensen P, Frandsen F, Glarborg P. Release to the gas phase of inorganic elements during wood combustion. Part 2: Influence of fuel composition. Energy Fuels 2008;22:1598–609. [33] Zevenhoven M, Skrifvars B, Yrjas P, Backman R, Muller C, Hupa M. A challenge for fuel characterization and modeling. In: 17th International conference on fluidized bed combustion, 2003. [34] Blander M, Pelton A. The inorganic chemistry of the combustion wheat straw. Biomass Bioenergy 1997;12(4):295–8. [35] Thy P, Jenkins B, Lesher C, Grundvig S. Compositional constrains on slag formation and potassium from rice straw blended wood fuel volatilization. Fuel Process Technol 2006;87:383–408.