Fly Ash Formation during Suspension Firing of Biomass: Effects of

Subscriber access provided by GAZI UNIV

Article

Fly Ash Formation during Suspension-Firing of Biomass. Effects of Residence Time and Fuel-Type Anne Juul Damoe, Peter Arendt Jensen, Flemming J. Frandsen, Hao Wu, and Peter Glarborg Energy Fuels, Just Accepted Manuscript • DOI: 10.1021/acs.energyfuels.6b02051 • Publication Date (Web): 23 Nov 2016 Downloaded from http://pubs.acs.org on December 7, 2016

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Energy & Fuels is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 57

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Fly Ash Formation during Suspension-Firing of Biomass. Effects of Residence Time and Fuel-Type. Anne Juul Damoe*; Peter Arendt Jensen; Flemming Jappe Frandsen; Hao Wu; Peter Glarborg. Department of Chemical and Biochemical Engineering, Technical University of Denmark, DK2800 Kgs Lyngby, Denmark

ABSTRACT

The objective of this work was to generate comprehensive data on the formation of residual fly ash during the initial stages of suspension-firing of biomass. Combustion experiments were carried out with pulverized biomass fuels (two straw fuels and two wood fuels), in an entrained flow reactor at 1200-1400 °C, simulating full-scale suspension-firing of biomass. By the use of a movable, cooled and quenched gas/particle sampling probe, samples were collected at different positions along the vertical axis in the reactor, corresponding to gas residence times ranging from 0.25 – 2.0s. The collected particles were subjected to various analyses, including char burnout level, particle size distribution, elemental composition, and particle morphology and composition. Furthermore, the transient release, i.e. the vaporization of the flame-volatile inorganic elements K, Cl and S, from the burning fuel particles to the gas phase, has been quantified by using two different calculation methods. The ash formation mechanisms were found to be quite similar for straw and wood. The degree of conversion (char burn-out level) was

ACS Paragon Plus Environment

1

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 57

generally good at residence times ≥ 1s. The size distribution of the residual fly ash particles evolved with residence time. For all ashes at long residence times a peak of residual ash particles in the range of 20 – 100 µm was observed. The residual ash particles were rich in Si, K and Ca. Further, at long residence times, submicron particles consisting primarily of KCl (condensed aerosols) became abundant in the ashes from straw combustion. Release of K to the gas phase was nearly 100 % for the two wood fuels and one of the straw fuels. A straw sample (Straw 2) with high Si/K molar ratio and a relative shortage of Ca showed a limited release of K in the range of 65 %; this suggests larger retention of K in Si-rich, Ca-lean fuels, due to incorporation of K into silicate structures. All S and Cl were nearly completely released to the gas phase for all studied samples.

1. INTRODUCTION: Replacing coal with biomass in existing suspension-fired power plants comprises an important aim in Danish Energy policy as well as globally on converting the energy supply system from largely fossil-fuel based, towards a system based on a large amount of renewable energy.1 However, a major obstacle for the conversion from coal to biomass in existing plants is to manage the ash behavior in such systems. Ash-related problems, such as increased deposit formation and boiler corrosion, and deactivation of SCR (Selective Catalytic Reduction) catalyst, are among the major technical challenges reported for suspension-firing of biomass.2-5 Ash-related problems are influenced by various physical and chemical transformations of the fuel during combustion. The route from a burning fuel particle in a furnace to ash deposit formation on a heat transfer surface can be divided into a number of consecutive steps:3,6-10

•

Conversion of the fuel by devolatilization, volatile oxidation, and char oxidation;


2

Page 3 of 57

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

•

Release of critical ash-forming elements to the gas phase, for biomass mainly K, Na, Cl, and S, during devolatilization and char burnout;

•

Formation of aerosol particles by nucleation and coagulation of flame-volatilized ashforming elements during cooling of the flue gas or by bursting of crystalline particles in the receding char structure (formation of CaO-rich aerosols);

•

Formation and entrainment of residual ash (condensed phase ash particles) during char burnout;

•

Transport of ash species, i.e., gases, liquids (droplets) and solids (particles), from bulk gas to heat transfer surfaces and adhesion of these ash species to heat transfer surfaces, and;

•

Build-up, sintering (consolidation) and shedding of deposits.

The fuel transformations such as devolatilization, char oxidation, fragmentation, and ash melting are among the important aspects of the ash formation processes in the radiant zone of a pulverized-fuel boiler. These transformations are both time-dependent and dependent on several fuel characteristics. In particular when it comes to biomass fuels, the transformations are presently not well described.3 Therefore, understanding of ash formation, and being able to predict the particle size distribution and chemical composition of the residual fly ash and the species release to the gas phase, when utilizing biomass as e.g. straw and wood for suspensionfiring, is important as a step towards minimizing ash–related problems in suspension-fired biomass boilers.


3

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 57

Through 50+ years, research has been conducted in order to characterize ash and deposit formation in utility boilers fired with coal, biomass and waste fractions. The basic mechanism of fly ash formation in pulverized fuel (PF) fired coal boilers is well described and may even be modeled relatively precisely.6-7,9,11 Concerning fly ash formation from biomass or waste used in suspension fired boilers, we do not presently have a full overview of the ash formation process. Fly ash data are available from boiler measuring campaigns where different ash fractions, sometimes including also in-situ ash, have been collected and analyzed chemically.2,4-5,12 Thus, there is some mechanistic understanding of the fly ash formation in plants fired with biomass or waste with well controlled local conditions, either alone or in conjunction with coal.2,4 However data from dedicated studies of the physical size development, ash species transformation and transient changes in chemical composition, of fly ash from pulverized biomass firing are limited, with Korbee and co-workers being pioneers in the field.3,13 Korbee et al.3 studied initial ash transformations of different coal and biomass fuels under typical PF-firing conditions by use of a Lab-Scale Combustion Simulator (LCS). Gas phase ash release, conversion, ash size reduction, and size distribution were derived alongside with changes in mineral chemical composition for different conversion levels at 20, 90, 210, and 1300 ms of residence times, and char burn-out, devolatilization level and fragmentations were quantified. The study included wood chips, waste wood, olive residue, straw, a UK coal, and a Polish coal, all having particle size < 1000 µm. A cooled and quenched gas/particle probe was used for extractive sampling of gas, char and ash at four locations along the LCS reactor vertical axis. Solid residue (char and ash) was sampled by a cascade impactor to obtain 11 fractions in the size range > 50 µm down to approximately 0.3 µm, and SEM-EDS was subsequently used to analyze each stage of the impactor. The inorganic matter released from fuel particles during pyrolysis


4

Page 5 of 57

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

and char combustion was then determined as the difference between the amount of inorganic matter in the fuel and the amount of inorganic matter left over in the solid residue after (partial) conversion. Particulate matter with a particle size smaller than 1 µm (aerosols) was mathematically added to the released part, as it was assumed that particles < 1 µm represents a fraction of the ash that has been released/volatilized during the combustion (aerosols), while coarse fly ash particles > 1 µm represent the residual (non-volatilized) fly ash.3 Ash recovery (not to be confused with probe sampling efficiency) was calculated by assuming the release of stable “marker” elements (such as Si) as zero. Korbee et al.3 found that all the studied biomasses, char was converted to a higher degree than for the coals. They found that the volatilization of inorganic matter (such as K, Cl, S, Ca, Si) was a time dependent process. Most of the inorganic matter volatilization took place in the first ~200 ms of the fuel conversion process; however, the release in the late burnout stage (1300 ms) was still sizeable. Large differences were observed between fuels. The release from a fuel is influenced by the fuel’s ash content and the reactivity of the ash constituents. As such, the total ash release from wood fuels was relatively low (< 5000 mg/kg fuel), reflecting a low ash content, while a high ash release from olive residue (~28,000 mg/kg fuel) and straw (~18,000 mg/kg fuel) was caused by the higher inorganic content in combination with a high ash volatility. The release from the coals was dominated by the elements sulfur and chlorine and amounted to 8000 mg/kg fuel for Polish coal and around 19,000 mg/kg fuel for UK coal. During the initial char burnout (at 20 and 90 ms), a significant increase in the concentration of aerosol particles (< 1µm) was observed, and it was interpretated to be a sign of burning with attritive fragmentation.3 Furthermore, fine (1 – 10 µm size) particle concentrations were found to decrease for all fuels after 90 ms, suggesting that that the lower size particles devolatilize and


5

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 57

oxidize quickly as compared to larger particles. The faster conversion of the smaller char particles was observed for all fuels except the Polish coal with the highest ash content. It was further suggested that after a certain conversion, larger (> 10 µm) particles fragmented more than the smaller particles and therefore their concentration decreased more rapidly in the later time steps, and biomass was found to be fragmenting more than coal. The elemental distribution of particles collected with different PSD was derived for each char burnout stage, and it was observed that S and Cl started to vaporize already at 20 ms in the flame itself, where the release of alkali and other minerals was still negligible. The alkali metals appeared to be vaporizing initially at the time step around 90 ms.3 In Si, or Si and Al rich fuels, such as straw and the two coals, the overall release of alkali minerals was limited (< 50 %). The release of Ca and Mg was significant in Ca and Mg-rich woody fuels (≥ 30 % vaporization). Alkali-rich fuel such as olive residue was found to be the most volatile compared to all other fuels.3 Shah et al.13 attempted to describe the complex elemental volatilization process under pulverized-fuel combustion conditions by means of simple linear correlations as a function of mineral matter compositions and their association in the char matrix. The release from 6 diverse biomass fuels and two different coals was studied experimentally in a Laboratory Scale Combustion Simulator, using similar test conditions (and fuels) as in the related study by Korbee et al.3 The raw release data were plotted against several chemical indices, in order to evaluate the effect of mineral matter compositions, i.e. Cl, Al, Si and S, onto the release of each individual element (K, Na, Cl, S, Ca, Mg). Linear regression lines were then plotted onto these charts, and in this way the elemental release was described as a set of linear correlations, presented for each element and for each fuel group (e.g. woody biomass, agricultural biomass/residue, coal).


6

Page 7 of 57

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

In this way, Shah et al.13 found a close match (R2 ~0.95) when plotting the release of K against the ratio of (K + Cl)/(Si + Al + 2S) in the fuel, implying that a high fraction of Si, Al and S reduces the fraction of K released, while a high fraction of Cl increases the K release. Almost complete release of K was observed for woody biomass with low contents of Al and Si. A close match with R2 > 0.99 was observed for S released against fuel levels, for various fuels (both biomass and coal), meaning that the relative fraction of S released was relatively constant and the absolute S release was proportional with the fuel level.13 Cl was found to be released completely from woody biomasses, while other fuels such as saw dust, olive residue, straw, and Polish coal, in some cases demonstrated noticeably lower release level. Only a partial release (up to a few %) of Ca and Mg was observed throughout the fuel range, although with a significant release (≥ 30 %) for woody biomasses containing very high shares of Ca in its ash. Fuels richer in Si, such as saw dust, olive residue and straw, released much less Ca.13 Other studies have also addressed the gas phase release of critical ash-forming elements such as alkalis, S and Cl from biomass fuels, or classes of biomass fuels, but primarily for fixedbed/grate-firing conditions.14-20 In these studies it is generally find that a high fraction of Cl enhances K vaporization due to the formation of volatile KCl, while Si reduces K volatility by incorporating K into silicate structures. The presence of Cl favors the formation of KCl over reactions between K and silicates.14-15,21 Knudsen and co-workers14-15 found from lab-scale release experiments with annual biomass fuels that it was possible to drive virtually all K into the gas phase (as KCl or (KCl)2), by addition of HCl to the fuel sample at temperatures above 900 °C. On the other hand, excess levels of Si and Al may inhibit the release of K, due to incorporation into stable K-Al-silicates.13-15


7

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 57

At fixed bed combustion conditions a large fraction of Cl is released as HCl or CH3Cl below 500 °C,22 and residual Cl is released as KCl at temperatures below 900 °C. Cl effect on alkali release is pronounced.13-14 S is to a high extent released to the gas phase, primarily as SO2. Further, SO2 may react with alkali chlorides to form alkali sulfates, if the fuel S/Cl molar ratio is high, or with Ca and Mg to form sulfates at < 1450 °C. Sulfur and aluminosilicates compete for reaction with alkalis. High temperature favors alkalis alumino-silicates over sulfate formation.13 Few studies address the effects of Ca and Mg on the behavior of Cl and alkali metals. Alumino-silicates are more likely to react with Ca and Mg than with alkalis. It is expected that higher levels of Ca will cause more alkalis to remain as gaseous alkali chlorides, sulfates or (hydr)oxides even at high temperatures.13-14 This effect has been confirmed by means of release experiments on synthetic samples of K-Ca-Si mixtures by Novakovic et al.21 Despite the numerous studies addressing release to the gas phase of ash forming elements from various biomass fuels, attempts to make a more general model for the release, applicable for a wider range of fuels, have been scarce.13-15 Knudsen and co-workers14-15 studied the release from annual biomass during fixed-bed combustion conditions and proposed simple, linear correlations for the release of K, Cl and S from annual biomass fuels, based solely on the inorganic composition of the fuels. The K release was found to depend in particular on the Cl and Sicontent in the fuel, and thus the fuels were initially categorized into two different categories, i.e. Si-lean fuels (K/Si molar ratio > 2), and Si-rich fuels (K/Si molar ratio < 2), exhibiting different release behavior. Release estimates for K included simple, linear expressions for min and max release, based on Cl/K and (Ca+K)/ Si molar ratios, were derived. Contrary to Shah et al.13 and others, Knudsen and co-workers14-15 also incorporated the liberating effect of a high Ca-content on the release of K from Si-rich fuels, in the expressions.


8

Page 9 of 57

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

The objective of the present work has been to characterize experimentally the formation of fly ash from suspension-firing of biomass. The study includes 1) ash transformations, i.e. changes in particle size distribution and chemical composition, and release of inorganic elements, as a function of residence time, and; 2) influence of fuel type and fuel ash composition on fly ash properties. This has been done by conducting combustion experiments with bio-dust in an entrained flow reactor. With the data obtained, we aim at complementing and extending the limited data available from previous studies in the field.3,13 It should be noticed that our study differs significantly from the studies by Korbee and co-workers3,13 on several parameters, most importantly the experimental set-up (and fuels) used, the conversion levels (residence times) studied, and the approaches we make for quantifying the release, as described in detail in the experimental section. 2. EXPERIMENTAL: 2.1. The entrained flow reactor: The experimental set-up consists of an electrically heated solid fuel entrained flow reactor (EFR) and additional support equipment, as shown in Figure 1. The reactor system basically consists of three parts: a fuel feeding and gas preheater section in the top; the reactor, consisting of a 2 m long ceramic tube with an inner diameter of 8 cm, lined with 7 heating elements (T max = 1500 °C); and a water cooled bottom chamber. The bottom chamber has a hole in the bottom for insertion of a gas/particle extraction probe, and several flanged openings in the side, of which one is connected to the exhaust system. The EFR has previously been used to study ash behavior during e.g. co-firing of coal and straw,2 and co-firing of coal and solid recovered fuel.4 For the present experiments, the setup was used to combust pulverized biomass fuels under conditions resembling a pulverized fuel furnace. The setup was


9

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 57

equipped with a movable gas/particle sampling probe, allowing collection of gas and particle samples at different gas residence times. 2.2. The gas and particle sampling system: The movable gas and particle sampling probe has been designed to extract and quench a fraction of the flue gas and particles, at different gas residence times (ranging from ~0.25 to ~2s) in the reactor. The movable sampling probe is water-cooled and thermo stated to 60 °C, in order to minimize water condensation in the probe. The gas and particle sample is quenched with N2 directly in the probe tip, in order to force nucleation of vaporized inorganic species. The quench N2 flow and suction flow can be adjusted in order to obtain isokinetic sampling conditions. The gas composition (O2, CO2, CO, NO, SO2) of the quenched flue gas from the probe and the raw flue gas from the reactor outlet is monitored continuously by standard gas analyzers, and the difference in CO2 concentration is used to determine the actual quench dilution ratio. The quenched and cooled sample gas from the probe is led to a particle sampling panel (see Figure 1), where it passes first through a cyclone which separates out the coarse fly ash fraction (cut-off diameter ~1.5 µm). From the cyclone, a fraction of the flue gas is directed through a 0.1 µm poly ethylene filter where the sub-micron aerosol particles are collected. The temperature in the extraction line, from the outlet of the probe and through the cyclone and aerosol filter, is maintained at ~80 °C to ensure that the flue gas temperature is above the water dew point. 2.3. Residual vs. released ash: “Released ash” is here defined as the vaporization of inorganic, ash forming matter (e.g. K, Cl, S) from the fuel during combustion. The released inorganic material may subsequently form submicron aerosol particles by nucleation and coagulation of the flame-volatilized (or burst) particles, or stay in the gas phase as e.g. HCl.


10

Page 11 of 57

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

“Residual ash” is defined as condensed phase (non-volatilized) ash particles, generally measuring > 1 µm. Similar to the assumption used by Korbee et al.3 we assume that the fraction of ash material captured in the aerosol filter (< ~1.5 µm) represents the aerosol fraction of the ash, i.e. it has been released (vaporized + condensed) during the combustion process. Scanning Electron Microscopy (SEM-EDS) examination of the collected aerosol filter material has confirmed that it consists of (agglomerates of) primary particles ( ~1.5 µm, similar to the definition by Korbee et al.3 However, examination of the particles separated in the cyclone by SEM-EDS revealed that the cyclone material contains both coarse fly ash particles representing the residual (nonvolatilized) fly ash and char fragments, in addition to a certain fraction of submicron particles rich in K and Cl, which in SEM appears to be aerosol particles deposited from the gas phase. The aerosol particles may be attached to the surface of larger fly ash/char particles, or be present as agglomerates larger than ~1.5 µm (for SEM pictures, please refer to Figures 6 and 7; the SEMEDS analysis results are provided as Supporting Information). This suggests that the fraction of submicron particles trapped in the cyclone represents part of the released ash fraction (as aerosols), and needs to be quantified separately. As a means to quantify the fraction of aerosol particles trapped in the cyclone, we assume that the aerosol fraction appears as salts (KCl and K2SO4) that can be determined as the fraction of water soluble K, Cl and S.2 This water-


11

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 57

solubility assumption, however, may overestimate the fraction of aerosols, as discussed further below. To further complicate the picture, in addition to the solid material collected in the cyclone and aerosol filter, material fractions are also deposited elsewhere in the sampling system, such as in the sample line (hoses) and on the probe top; these fractions also need to be considered with respect to the overall mass balances and release quantification, as discussed in detail in the following section. 2.4. Mass balances and quantification of release: A schematic overview of the mass flows (or fractions, denoted by the coefficient ή) and characterization routes (residual vs. released ash) for the measurements is shown in Figure 2. After each experiment, the solid material (ash particles and possible unconverted fuel and char particles) collected in the cyclone and aerosol filter are weighed and saved for analyses. Additionally, three solid material samples are recovered and weighed: Material deposited in the sample line between the probe and the cyclone (“coarse ash in hoses/pipes”, fraction M3), material deposited on the probe top (“coarse ash on probe top”, fraction M2), and condensed salts deposited at the quench gas inlet inside the probe tip (“fine (condensed) ash in probe tip”, fraction M5). M1 (“not sampled fraction”), M4 (“not collected (coarse) particles”) and M6 (“not collected aerosol”) all represent solid material fractions that are “not recovered” (i.e. they are lost in the system). The losses may be attributed to accumulation/deposition on reactor wall, losses in the sampling lines, and losses due to direct emissions to the flue gas exhaust system of the fraction not sampled by the probe. By comparing the chemical and physical characteristics of the material collected in the cyclone (Mcy) and aerosol filter (Mfa) with spot check analysis of the additional material flows (M2, M3


12

Page 13 of 57

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

and M5), the assumptions a) – c) listed below were regarded to be reasonable, considering the accuracy of mass balances and quantification of release. The total mass collected from each experiment generally amounted to 0.5 – 1.0 g, with the cyclone material (Mcy) making up the largest fraction (often 30 – 80 %). The mass collected in the aerosol filter (Mfa) was very fuel dependent. In the case of straw fuels, due to the high volatility of straw ash, it was in the similar range as the mass collected in the cyclone. For wood fuels, on the other hand, the mass of the aerosol filter material generally amounted to less than 10 % of the mass collected in the cyclone, and it could be as low as < 0.05 g, which means that it was not always possible to do wet chemical analysis on these fractions. The additional mass fractions collected in the sample lines and probe (M2, M3, M5) varied from one experiment to the other, but the sum of these fractions was generally less than 30 – 40 % of the total mass collected. For ash and elemental mass balance calculations, the following is assumed for each experiment: a) The concentration of organic matter in the material collected in hoses/pipes (M3) is the same as the concentration of organic matter determined in the cyclone material (Mcy), while the concentration of organic matter in the aerosol filter material (Mfa), in the condensed salts in the probe tip (M5), and in the deposit on the probe top (M2) is assumed to be zero; b) Material collected on the probe top (M2) and material collected in hoses/pipes (M3) have an ash composition similar to material collected in the cyclone (after correcting for organic matter content according to assumption a);


13

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 57

c) Condensed salts inside the probe tip (M5) has a composition similar to the aerosols collected in the aerosol filter. It is further assumed that: d) All elements are to an equal degree deposited in the reactor and collected by the probe, respectively. For quantification of the release, it is further assumed that: e) The fraction of aerosol particles trapped in the cyclone appears as salts (KCl and K2SO4) that can be determined as the fraction of water soluble K, Cl and S (ήw). The water soluble part of K, Cl and S can therefore be anticipated to have been released from the fuel particle to the gas phase. Assumption e), however, is inaccurate if the sample contains significant amounts of unconverted fuel (char) particles, due to the fact that organically associated K, Cl and S will also be water soluble in most biomass fuels. Therefore, the water solubility assumption may overestimate the fraction of K, Cl and S that has actually been released, especially at the short residence times. Furthermore, KCl particles (and other salts) found on the surface of char particles may have been generated, and migrated as a liquid phase, in the pores and onto the surface of the char.23 Such particles, while water soluble, have not (yet) been released to the gas phase. In order to comply with this uncertainty, it may be relevant to operate with a “maximum release” and a “minimum release” constrain when estimating the gas phase release. The “minimum release” can in that case be estimated by defining the water soluble fractions of K, Cl,


14

Page 15 of 57

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

and S in the cyclone ash as not released (i.e. ήw ≡ 0). This, on the other hand, may be an underestimation of the actual release. According to the characterization route in Figure 2, and assumption a) above, the ash mass balance recovery, ήrecov,ash (recovered fraction in experiment compared to input fuel ash) is then defined as: 1 ή ,

1 , 1 − , + + ή + + = = , 1 − ,! ! ή

Here corg,cy is the mass concentration of organic matter measured in the material collected in the cyclone (e.g. 0.85 g/g), and corg,o is the mass concentration of organic matter in the fuel (e.g. 0.99 g/g). If no ash is lost in the reactor system, ήrecov,ash will be equal to 1. Further, when using assumption b), c) and d), the recovered fraction of an element becomes: 2 ή ,#$ = =

#$, #$,

% + +

1 1 & + % + & + 1 − , ή ! ! ή

When including assumption e), the elemental release, ήrel (method 1), calculated based on the sample recovered (Melement,out) in the experiment, is defined as:


15

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

3 ή# $ ( )

Page 16 of 57

1 1 & ή + + + 1 − , * ή = 1 1 % + + & + % + & + 1 − , ή % + +

Here ccy is the mass concentration (e.g. mg/kg, or g/g) of the element determined in the material collected in the cyclone, ήw is the water soluble fraction of the element in the cyclone material, cfa is the mass concentration of the element determined in the aerosol filter material, and cg is the concentration of the element determined in the flue gas (applicable for S only (as SO2)). The elemental release quantification method outlined in equation (3) is hereafter termed Release Quantification Method 1. The gas phase release Method 1 determination is solely based on the collected ash fractions (and flue gas SO2), Melement,out. An alternative quantification method, hereafter termed Release Quantification Method 2, is also introduced, based on a slightly modified version of the method suggested by Korbee and coworkers.3,13 The release using Method 2 is calculated as the difference between the amount of inorganic matter in the fuel (corrected for experimental ash recovery ήrecov(Ca,Si)) and the amount of inorganic matter left over in the (non water soluble) coarse ash fractions after (partial) conversion. c0 is the mass concentration of the element in the fuel (4): 4 ή#$ (

1 % + + & 1 − ή* 1 − , =1−, / ! ! ή ,- ,.


16

Page 17 of 57

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

The experimental recovery ήrecov(Ca,Si) (ash mass balance) for each experiment is calculated by assuming Ca and/or Si being stable “marker” elements exhibiting zero release (i.e. Ca and Si are used as ash tracers; assuming that these elements should be transferred 100 % to the coarse ash fractions) (5): 5 ή ,- ,. =

- ,., - ,.,

Compared to the method used by Korbee and co-workers,3 we have modified the Method 2 to include also the fraction of water-soluble K, Cl and S in the cyclone ash (ήw) in the “released” part of the ash, as seen from equation (4). I.e., the gas phase release Method 2 determination is based on using Si and Ca as an ash tracer and then assume that the non water leachable ash fraction has not been released. Each quantification method has advantages and drawbacks. The major advantage of quantification method 1 is that it is comprehensive in including all (5) collected ash fractions (in addition to the flue gas composition when applicable) in the calculations, as specified in Figure 2 and equation (1) – (3). By using Equation (1) the relative ash collection recovery efficiency can be tested. In most experiments an ash recovery between 45 and 98 % was obtained, (see Table 2 in the results section). This is considered acceptable, as previous experiences with the EFR indicate that the experimental ash recovery is often not greater than 40 - 80 % and deviates considerably between experiments, due to deposition on reactor wall and other surfaces.2,4 In a single experiment (Wood 2 1400 °C 1.0s) an ash recovery of 153 % was obtained; this was probably a consequence of analytical limitations (discussed below).


17

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 57

An important limitation of the Quantification Method 1 is related to the analytical determination of specific elements especially in the “released” (aerosol filter) ash fractions, where wet chemical analysis is not always possible, due to a too small (< 0.05 g) amount of sample mass available. Instead, the chemical composition in these cases is estimated from EDS analysis, which has limited accuracy because light elements (< Z=11) cannot be detected and the detection limit for the remaining elements is rather high (> 1-2 %). Furthermore, the release Quantification Method 1 prerequisites input data for both the fraction of sample gas that is directed through the aerosol filter (ήa), and the content of unburnt organic matter in the collected ash fractions (which is determined by STA (Simultaneous Thermal Analysis)). Neither of these input data are considered very accurate. The major advantage of Quantification Method 2 is that it is simple and does not require input data for aerosol ash composition. A general limitation for the quantification of the release of specific elements (for both methods) is that the concentration of the element in certain material fractions may be close to, or even below, the detection limit of the chemical analysis (this is in particular due for the Wood 2 fuel and ash fractions, and for the elements S, Si, Cl, Na). The gas phase release Quantification Method 1 (equation 3) determination is solely based on the collected fractions, and it is assumed that the water soluble fraction has been present in the gas phase. The gas phase release Quantification Method 2 (equation 4) determination is based on using Si and Ca as an ash tracer and then assume that the non water leachable fraction has not been released.


18

Page 19 of 57

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

2.5. Fuels utilized: Two straw fuels and two woody biomasses were used in the EFR experiments. Fuel analyses are provided in Table 1. All fuels were milled/pulverized, prior to use in the EFR (see Table 1 for milling procedure). The resulting particle size distribution (PSD) was measured by laser diffraction analysis and is listed as Dv(10), Dv(50) and Dv(90) values in Table 1. The two straw fuels are characterized by high contents of K, Si, Cl and Ca, and an ash content > 4 %. Straw 1 has significantly higher content of K, Cl and S as compared to Straw 2, while the content of Si is lower. The two wood fuels are dominated by Ca, K and Si, a low ash content ≤ 1 %; and only traces of Cl. Further, it is noticed that Straw 2 and Wood 2 (which were received as pellets and subsequently size reduced by using lab-scale milling and sieving equipment, see Table 1) contain a significant fraction of particles larger than ~700 µm according to the laser diffraction analysis. This is probably due to the presence of fiber structures (highly non-spherical particles) in these fuels after milling. The two other fuels, Straw 1 and Wood 1, had been pulverized by industrial milling and sieving equipment and contain less fiber structures. 2.6. Experimental matrix: An experimental matrix, testing two different reactor temperatures, and three different sampling positions/gas residence times, respectively, was set up (see details in Table 2). The three sample probe positions corresponded to gas residence times in the range 0.25 to 2.0 seconds, depending on the reactor temperature. In order to obtain an excess air ratio around 1.5 (corresponding to an exit flue gas O2 level of approximately 6.5 vol. %), the inlet air flow rates in the study were 30 NL/min, and the particle feed rates were in the order of 200 - 300 g/h. The duration of an experiment was limited by the pressure drop in the particle sampling system, i.e. the experiment continued as long as a constant flow could be obtained through the


19

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 57

aerosol filter (≤ 35 min in case of straw fuels, ≤ 90 min in case of wood fuels). Steady-state conditions (i.e. stable flow and temperature) were attained. 2.7. Characterization of ash particles: Selected samples of the collected cyclone and aerosol filter particles were subjected to the analytical methods outlined below: Characterization of cyclone particles by STA analysis. Simultaneous Thermal Analysis (STA) was employed in order to determine the content of organic matter of the cyclone ash material. A sample of 5 mg of cyclone material was loaded in a Platinum (straw ashes) or an Alumina (wood ashes) crucible and heated at 10 °C/min (in N2 and N2+O2 atmosphere) to the final setting temperature of 1200 °C in a thermogravimetric apparatus (Netzsch STA-449F1). The applied temperature program and gas environment can be found in Supporting Information. The content of organic matter in the cyclone material (corg,cy), as determined by the STA analysis, was then used to estimate the char burn-out level, as determined by the ash tracer method outlined in equation (6):4 5 6 2 = 31 − × )!!64

4

5

)!!648 48

9 × 100

Here B (%) is the char burn-out, A0 (%) is the ash content (mass percent) of the dry fuel (from Table 1), and Ai (%) is the mean ash content (mass percent) in the material collected in the cyclone. Particle size distribution of residual (cyclone) ash by laser diffraction. The particle size distributions of the pulverized fuels and residual (cyclone) ashes, respectively, were determined by laser diffraction using a Malvern Mastersizer, in order to investigate the particle size development of the residual fly ash as a function of residence time. All samples were dispersed


20

Page 21 of 57

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

in ethanol during the size measurements. As the method requires quite large amounts of sample mass (generally > 100 mg), only a limited number of cyclone ash samples could be analyzed. Wet chemical analysis of selected coarse and fine ash fractions. The chemical composition of selected coarse and fine ash material fractions (cyclone material (Mcy), aerosol filter material (Mfa), hoses/pipes material (M2)) were determined by wet chemical analysis (pressurized acid digestion of 50 mg sample and subsequent determination of various elements by ICP-OES analysis). The elements determined included Ca, Cl, K, Mg, Na, P, S, Si, and water soluble fractions of Cl, K, and S. The results obtained were used as input for the elemental mass balance calculations, and quantification of the release. However, as wet chemical analysis requires sample masses larger than approximately 50 mg, it was not always possible to analyze all relevant ash fractions, especially not all aerosol fractions. SEM-EDS analysis: Scanning Electron Microscopy-Energy Dispersive X-ray analysis was employed to study the morphology (structure, size, shape) and (semi-quantitative) elemental composition of selected coarse and fine ash fractions. The semi-quantitative elemental composition analysis by EDS was used as a supplement to the wet chemical analysis, as this was not always applicable (e.g. on aerosol particle fractions). 3. RESULTS AND DISCUSSION: This section focuses primarily on the properties (chemical composition, experimental recovery, PSD, etc.) of the residual (cyclone) ashes obtained at the different experimental conditions. The concentrations of inorganic elements (K, S, Cl, Ca, Mg, P, Si) in the materials collected by the cyclone as a function of ash content, is plotted in Figure 3. Ash content and experimental ash recovery for each experimental condition are provided in Table 2, and char burnout as a function


21

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 22 of 57

of residence time are presented in Figure 4. Characteristic particle size distributions as a function of residence time is plotted in Figure 5. For comparison with the raw fuel properties, please refer to Table 1. 3.1. Ash chemical composition and char burn-out: Figure 3 reveals that the concentration of inorganic elements in the cyclone material, including the concentration of the more volatile elements K, Cl, and S, generally increases with increasing ash content. As a consequence the concentration of water soluble K, Cl and S in the cyclone material increases. Furthermore, it is noticed from Table 2 that the ash content increases with residence time, even though the content of unburnt organic matter in the cyclone material is still significant even at 1.7 – 2.0s residence time (up to 70 %), implying that the char burnout is not completed. From Figure 4 it is seen that the char burnout level is time and fuel dependent, while less dependent on temperature. The two wood fuels reach lower overall degree of conversion in the first ~1s of the combustion process, as compared to the two straw fuels. However, it should be noticed that the different fuels may not be directly comparable, as e.g. the particle size and shapes vary significantly. The char burnout is generally good (> 98 %) at residence times > 1s. 3.2. Particle size development and chemical composition of residual fly ash: Characteristic particle size distributions for the raw fuel and the residual (cyclone) fly ash collected at different gas residence times are depicted for a number of experiments in Figure 5. SEM pictures illustrating the morphology and particle size of the cyclone ash particles at different levels of conversion (at 1400 °C) are provided in Figures 6 and 7. The discussion below of the particle size development of the residual fly ash is based on the PSD measurements and the corresponding SEM analyses.


22

Page 23 of 57

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Straw fuels, 1200 and 1400 °C: The measured size distributions (PSDs) of the Straw 1 fuel exhibits a broad peak with a maximum at ~400 µm (Figure 5 a) and c)). For the Straw 2 fuel (Figure 5 e)), the size distribution function involves a broad peak with a maximum around 650 µm. With increasing degree of conversion, the PSD of the cyclone ashes changes towards a bimodal particle size distribution: (1) a coarse-mode peak appearing at dp > 300 µm, and (2) an intermediate-sized peak with max dp around 30 – 50 µm. Further, a fine-mode peak, which is considered to be aerosols formed from condensation of vaporized species (K, Cl and S), is occurring at dp < 1 µm (not shown on Figure 5). The fine-mode peak is not considered part of the coarse ash, and this peak is thus not discussed further here. The coarse-mode peak (1) possibly consists of particles with some degree of unburnt organic material, while the intermediate-sized peak is attributed to mineral coalescence and fragmentation of the char particles, as discussed in more detail below. The evolution in PSD seems to be consistent for Straw 1 ash converted at 1200 and 1400 °C, respectively. At the initial 0.25 – 0.28s of conversion of Straw 1 at 1200 and 1400 °C, the fuel peak size decreases. The shift towards slightly smaller particle sizes could be caused by fuel conversion and char fragmentation. The shape and position of the peak, as well as a significant content of organic matter of about 70 % in the collected material, indicate that the major contributor to the PSD at 0.25 – 0.28s is fuel/char particles that are only partly converted. SEM investigations of the Straw 1 1400 °C 0.25s ash (Figure 6 a)) confirms that the sample contains numerous large, porous, partly converted fuel particles, measuring up to >500 µm in diameter. At 1.0s residence time, for Straw 1, the broad peak of large fuel/char particles originally positioned around ~400 µm is narrowed and shifted towards significantly larger particle sizes (~600 µm). The relative increase in particle size may suggest agglomeration of char/ash


23

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 24 of 57

particles. SEM-investigations of the Straw 1 1400 °C 1.0s ash (Figure 6 b)), however, show no clear evidence of agglomeration. Two characteristic, large (~500-600 µm long) particles covered with numerous tiny particles are seen in the right-hand side of Figure 6 b). This picture indicates coalescence of ash on receding char surfaces. The tiny particles covering the char surface at this stage mainly contain K and Cl, most probably KCl particles (SEM-EDS analysis is provided in Supporting Information). The KCl particles may have migrated as a liquid phase in the pores and onto the surface of the char23 or condensed from the gas phase salt species on the char particles during cooling. The content of unburnt organic matter in the Straw 1 1400 °C 1.0s ash is still significant (27 %), emphasizing that char particles perhaps remain the largest contributor to the large-particle size mode positioned at ~600 µm. The apparent shift towards larger particles at 1.0s, as compared to the initial fuel PSD, may simply reflect the relatively faster conversion of the smaller fuel particle sizes as compared to the larger particles. At residence times ≥1.0s, an intermediate-sized peak with maximum around 45 µm develops with increasing residence time/degree of conversion, while the concentration of larger particles above 200 – 300 µm is fading. SEM-EDS investigations of the Straw 1 1400 °C 1.0s and 1.7s ash (Figure 6 b) and c) and Supporting Information) reveal that the particles in the intermediate size interval from around 10 – 100 µm is predominantly a mix of spherical or irregular fly ash particles rich in Si, K and Ca, char fragments, and aggregates of submicron particles composed largely of KCl; the general appearance seems to suggest that the submicron particles are aerosol particles deposited from the gas phase. The aggregates of submicron aerosols are often (partly) covering the fly ash and char particles. The unburnt organic matter content in the Straw 1 ash at 1.7 s / 2.0s is 10 – 12 %, indicating that some further development of the PSD may occur at even longer residence times.


24

Page 25 of 57

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

The SEM-EDS investigations of the Straw 2 1400 °C 1.7s ash (2 % organic matter) (Figure 6 d) and Supporting Information) reveal numerous, almost perfectly spherical fly ash particles measuring around 10 – 100 µm in size. Beside these spherical particles, we observe a pozzolanic structure (particle with high Si content); characteristics of an almost fully developed fly ash. KCl-rich, submicron aerosols that appear to be deposited from the gas phase are less abundant in the Straw 2 ash, as compared to the Straw 1 ash, reflecting the lower fuel concentration of K in Straw 2, and hence the lower total ash vaporization (Table 1). Wood fuels, 1200 and 1400 °C: The size distribution of the original Wood 1 fuel presents a broad peak in the size interval 10 – 700 µm, with a maximum at ~500 µm. For the Wood 2 fuel, the size distribution function presents a broad peak with a maximum around 650 µm. As for the straw fuels, the physical size distribution of the char/ash evolves with residence time, towards a bi-modal size distribution. At the initial 0.25s of conversion, the Wood 1 1400 °C ash particle concentration in the size interval 50 – 240 µm is decreased, while a relative increase in the concentration of larger particles (240 – 700 µm) is seen. This indicates that smaller fuel particle sizes convert (devolatilize and oxidize) quickly compared to larger particles, consistent with the observations by Korbee et al.3 The content of organic matter in the Wood 1 1400 °C ash at 0.25s residence time is still very significant (97 %), implying that the fuel is only partially converted and suggesting that the larger particles peak consists primarily of slightly converted fuel/char particles. The char structures appear in the SEM (backscattered) electron images as larger (10 – >500 µm), quite dark, irregular particles with a porous structure (Figure 7 a)).


25

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 26 of 57

At ≥ 1.0s residence time, the concentration of the larger particles above 240 µm is reduced with increased conversion of the Wood 1 char, while a new peak with a maximum around 30 – 50 µm develops. The evolution in PSD seems to be consistent for ash converted at 1200 and 1400 °C. SEM-EDS investigations of the Wood 1 1400 °C 1.0s and 1.7s ashes (Figure 7 b) and c) and Supporting Information) reveal that the particles in the intermediate size interval from around 10 – 100 µm consist of spherical fly ash particles rich in Ca and Si, with varying content of K, in addition to irregular ash and char particles. The surface of the (receding) char particles may be irregular, with some of the original wood structure preserved, or smoothed/molten, and to some extent covered with smaller ash particles/droplets. Submicron aerosols that appear to be deposited from the gas phase are less common in the Wood 1 ashes as compared to the ashes from the two straw fuels; this directly reflects the lower concentration of alkali and Cl in the wood as compared to the straw, and hence the lower total ash release (Table 1). The content of organic matter in the particles is still significant even after 1.7 s residence time (19 %), suggesting that further development of the PSD may occur with longer residence times. For the Wood 2, only limited PSD data are available for the converted particles (1.7 s); however the major tendencies seem to be similar as for the other wood sample (Wood 1): With increasing time of conversion, the particles shift towards a bi-modal particle size distribution and smaller particles. The SEM picture of the particles at 1.7 s of conversion (Figure 7 d)) reveals that the Wood 2 particles mainly consists of porous char particles/fragments and pozzolanic structures (particles with high Ca and Si content), measuring up to more than 100 µm in diameter and being partly covered by condensation of Na, S, Cl, K compounds. This is consistent with the fact that the Wood 2 particles are the least converted of all the fuels, containing 70 % of organic


26

Page 27 of 57

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

matter (30 % ash) at 1.7 s of conversion. The surface of the large char particles looks porous and partly molten, with some small bright particles on the surface. Straw ashes vs. wood ashes: In conclusion, the present PSD and SEM-EDS-analyses of the fly ashes suggest fragmentation of char and mineral inclusions as well as melting of ash as primary fly ash formation mechanisms, for both the straw fuels and the wood fuels. With increasing degree of conversion (i.e. at residence times ≥ 1s), an intermediate-sized fragmentation peak with peak maximum in the range 30 – 50 µm is formed. Furthermore, the fly ash fraction within the size interval [10 – 100 µm] appears to be quite similar for the straw and wood fuels, in the sense that it is dominated by molten or partially molten ash particles rich in Si, K and Ca. The fully converted fly ash particles from wood primarily consist of Ca and Si, with varying amounts of K, while the fly ashes from straw are rich in Si and K with less Ca, reflecting the differences in the original fuel chemistry. The fly ashes from wood fuels contain significant amounts of unburned organic material, even after 1.7 – 2.0s of conversion, and these appear in the samples as partially converted (char) particles with some of the original wood structure preserved. A further analysis of the fly ash PSD, including a model for predicting the degree of fragmentation of the fuel particles is provided by Hansen.24-26 In her modeling work it is predicted that when vaporized species are excluded, the degree of fragments formed per fuel particle from both Wood 1 and Straw 1 is just above 3, i.e. in the same range as those calculated for lignite by Sarofim et al.27 3.3 Quantification of release: Ash and elemental mass balances were established for all experiments, according to the method described in equation (1) and (2). For each experiment and across fuels, the ash mass balance (ήrecov,ash) generally closed at > 43 % recovery, as referred in


27

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 28 of 57

Table 2. Less than 100 % ash recovery indicates a loss of particles – most probably due to ash deposition on the reactor walls. More than 100 % recovery (153 % for Wood 2 1.0s, see Table 2) is explained by relatively large analytical uncertainties as a consequence of a very low ash content in the fuel. The corresponding elemental recovery (ήrecov,element), calculated by the two methods (total ash balance, or Ca/Si tracer balance, respectively) generally varied between 50 – 200 % for K, 90 – 260 % for Cl, 30 – 150 % for S, 30 – 160 % for P, and 40 – 160 % for Ca/Si, with a few lower or higher outliers (elemental balance results are provided as Supporting Information). As for the ash mass balances, the outliers are explained by large analytical uncertainties because the concentration of some elements are close to (or even below) the analytical detection limits. In particular, data for Wood 2 contained numerous outliers. It is noticed that the elemental mass balances for Cl, and to some extent also for K, have a tendency to close at more than 100 %. This can probably be tracked back to a relative uncertainty on the aerosol mass balance. The aerosols contain high concentrations of Cl and K, so a small overestimation on the aerosol mass has significant influence on the elemental mass balances of Cl and K. Now, while defining the elemental release according to equation (3) (Method 1) and equation (4) (Method 2), and taking into account the “minimum” and “maximum” release constrains for K, Cl and S discussed under assumption e), the elemental release of K, Cl and S is plotted as a function of residence time in Figures 8 - 10. Release data are shown only for volatile ash forming elements that are present in a reasonably high concentration in the fuel, as otherwise the uncertainties on the results are too high. In addition to the release of the volatile elements K, Cl and S, also phosphorous release may be important for some biomass fuels. Release data for


28

Page 29 of 57

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

phosphorous are included as Supporting Information, and the results are briefly discussed at the end of this section. Release method 1 (Equation 3) (referred to as M1 in the legends in Fig. 8 - 10) is based on the collected products, such that the released fraction is related to the fraction in the collected products that have been present in the gas phase (aerosol particles and gas phase). Release method 2 (Equation 4) (referred to as M2 in the legends in Fig. 8 – 10) uses an inert tracer (Ca or Si) and is based on the relative amount of a non-volatile fraction of the element that is collected. That is then assumed to be the fraction that is not released. When calculating the “minimum” release the cyclone ash water soluble elements are assumed not to have been released (i.e. ήw ≡ 0 in Eq. 3 and 4). When calculating the “maximum” release, the cyclone ash water soluble elements are assumed to have been released. The input values for water soluble fractions of K, Cl and S, ήw, in cyclone ashes that have been used in the calculations are provided in Table 3. The values listed are the average of several measurements. We believe that at least at longer residence times (≥ 1s) the water-soluble species found in the cyclone ash are probably from recondensation of gas phase salt species during sample cooling; and thus the water-soluble fraction provides a reasonable estimate of the released species. In order to check the assumption that Ca and Si can be used as “stable tracer elements” for the Quantification Method 2 calculations, the release of Ca and Si was also quantified by Method 1. This was done as “minimum” release quantification, i.e. the release calculation included Ca and Si recovered in the aerosol fraction, but any water soluble fractions were excluded (as they were not determined). These calculations show that the “minimum” release of Ca and Si was generally


29

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 30 of 57

less than 5 - 10 %, supporting the accuracy in using Ca and Si recovery % as basis for the elemental mass balance calculations (“stable marker elements” assumption) when using Method 2. Release of K: We believe that the max release data probably provide the best estimates of the potassium release. At very short residence times the minimum release could provide reasonable data; but at higher residence times ≥ 1s the water soluble species found in the cyclone ash are probably from recondensation of gas phase salt species during sample cooling. The two quantification methods provide reasonably similar data. Figure 8 reveals that the release of K is high throughout the tested range of fuels and residence times, though with some noticeable differences between the fuels. Depending on the quantification method used, the “max” release of K from Straw 1 and Wood 1 amounts to 70 – 98 %, increasing slightly with increasing residence time and/or temperature. For the “min” release, the residence-time dependence is pronounced for Straw 1 and Wood 1, increasing from around 35 % in average at 0.25 – 0.28s, to about 70 % at 1.7 – 2.0s in case of Straw 1, and from about 40 % to around 65 % in case of Wood 1. The Straw 2 shows significantly lower release as compared with the other fuels. The “max” release amounts to approximately 70 % and is not really dependent on residence time, and the “min” release never exceeds 60 %. The relatively low K release from Straw 2 may be attributed to differences in mineral matter compositions between the fuels, such as levels of silicon relative to K and Ca (see Table 1). According to the definition by Knudsen and co-workers,14,15 Straw 2 is a Si-rich fuel ((K/Si) = 0.5), with relative shortage of Ca: ((K+Si)/(Ca+Mg) = 5.2), and as such larger retention of K in silicate structures may be expected.


30

Page 31 of 57

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

The generally large release of K already at the initial 0.25 – 0.28s of conversion is in good agreement with the results obtained by Korbee et al.3 who found in their experiments that the alkali metals appeared to be vaporizing after about 90 ms of conversion. The results for Wood 2 are scattered with no clear trends. We attribute this to large uncertainties in the chemical analyses as a consequence of low concentrations of K. Release of Cl: Cl can be released not only as salts (e.g. KCl, NaCl) that are subsequently recovered in the aerosol fractions, but also as permanent gas as HCl(g) or CH3Cl(g).14,22 The distribution of Cl among KCl and HCl is determined by the equilibrium of HCl/KCl, which, again, depends upon the local stoichiometry and temperature and can furthermore be influenced by the speciation of K.28 Excess of K as compared to Cl (and S), and low temperatures, favors the formation of KCl. In our measurements, HCl(g) in the flue gas has not been determined. Both release methods calculate the total release of Cl (including both KCl and HCl release). Figure 9 depicts the release pattern of Cl for the two straw fuels. Data for the two wood fuels are not shown, as the results are scattered and uncertain due to low Cl concentrations in the residues. The “max” release of Cl from the straw fuels is close to complete and independent of residence time and temperature. (This is also the general case for the wood fuels, although the results are scattered). Furthermore, considering the “min” release (i.e. ήw ≡ 0), the trend seems to be similar as for K. In the case of Straw 1 the “min” Cl release increases from around 50 % at 0.25 - 0.28s, to around 75 % at 1.7-2.0s. For Straw 2, the “min” Cl release is around 65 %. It is noticed that the Cl release (both “min” and “max”) calculated by Method 2 (accounting for gaseous Cl), is generally comparable to the release according to Method 1 (neglecting gaseous Cl). This indicates that the fraction of Cl that is released from the fuel as permanent gas (e.g.


31

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 32 of 57

HCl(g)), is small, as compared to the fraction of particle bound Cl. If the fraction of gaseous Cl was significant, Method 2 should exhibit a higher release as compared to Method 1, which is not the case. Release of S: Figure 10 depicts the release pattern for S for the two straw fuels. Again, data for the two wood fuels are not shown, as the results are scattered/uncertain due to low S concentrations in the residues. The release of S exhibits similar trends as for release of Cl. The “max” release of S is close to complete for the depicted straw fuels, independent on residence time; while the “min” release exhibits an increasing trend with increasing residence time. For Straw 1, the “min” S release increases quite steeply from around 45 in average at 0.25 - 0.28s to around 80 % at 1.7 - 2.0s. For Straw 2, the “min” S release increases from around 65 % at 1.0s residence time to around 75 % at 1.7 s. It should be noticed that the SO2 concentration measured in the flue gas from the probe was very low/insignificant throughout the experiments, indicating that SO2 initially released to the gas phase may have reacted to form alkali sulfates and/ or Ca/Mg sulfates at the points of sampling.13 As for gaseous Cl, a significant fraction of permanent gas (SO2) “missing” in the Method 1 calculation would mean that calculation Method 2 should present generally higher release as compared to Method 1, which is not the case. Considering the release of P (results provided as Supporting Information), the release is negligible for the wood fuels while the straw fuels show significant, and temperature dependent, release. The P release from Straw 1 is about 20 % at 1200 °C / 2.0s, but more than 80 % at 1400 °C / 1.7s. 4. CONCLUSIONS:


32

Page 33 of 57

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

The focus in the present study has been to investigate the initial stages of fly ash formation during biomass combustion at suspension-firing conditions through combustion experiments in an entrained flow reactor above ~1000 °C. Ash transformations, char burnout and physical size development of residual fly ash, and transient release of flame-volatilized inorganic species, were studied. The following conclusions can be drawn: The ash formation mechanisms are found to be quite similar for both the straw fuels and wood fuels in the sense that:

•

Within the initial 0.25 – 0.28s after the fuels are injected into the hot zone (≥ 1200 °C) of the reactor, the flame volatile inorganic elements, K, Cl and S, are released to the gas phase to a significant extent. The degree of char burn-out at this point is in the range 72 – 91 %. Some of the smaller sized fuel particles have been fully converted to ash at this point, but char particles with a size ≥ 50µm, and with a significant content of organic matter, is by far the largest contributor to the PSD.

•

At 1.0 - 1.1s of residence time, char burn out, mineral particle fragmentation and ash particle melting have taken place. This is reflected in the PSD as the evolution of a new, intermediate-sized (fragmentation-mode) peak with maximum dp around 30 – 50 µm, while the concentration of larger particles above 200 – 300 µm is fading. The particles in the intermediate-sized interval appear in SEM as a mixture of spherical and irregular fly ash particles, and some char fragments. The char and ash particles may be more or less covered with smaller ash particles/droplets, originating partly from coalescence of ash droplets on the receding char surface and partly from condensation of salt species on the particles during the cooling process.


33

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

•

Page 34 of 57

At 1.7 – 2.0s of residence time, the PSD is further developed. However, in most cases some organic matter in the ashes at 1.7 s / 2.0s is still present, indicating that further development of the PSD may occur with longer residence times. In the wood ashes in particular, char particles with some of the original wood structure preserved are abundant, even after 1.7 – 2.0s of conversion.

After cooling of the flue gas, submicron aerosol particles, consisting primarily of KCl, are now abundant in the ashes from straw combustion, and aggregates of the submicron aerosol particles are covering the surface of several of the larger fly ash particles. For all ashes at longer residence times were observed a peak of ash particles in the range of 20 to 100 µm. Release of K to the gas phase were nearly 100 % for Straw 1 and the wood samples. A straw sample with high Si/K content (Straw 2) showed a release of K in the range of 65 %. All S and Cl nearly completely released to the gas phase for all studied samples. Based on the herein obtained results and the existing literature in the field, we may suggest the major transformation routes and transformation processes for ash forming elements during biomass suspension-firing as depicted in Figure 11. For practical applications, the experimental data obtained in this study suggest that the release of K is to some extent temperature dependent in the studied temperature interval 1200 °C – 1400 °C, implying that limiting the combustion temperature could reduce the release of K, which would decrease the formation of alkali chlorides and sulfates by gas phase reactions. This in turn would limit the quantity of aerosols and the extent of deposit formation and corrosion inside the combustion unit. On the other hand, while the release of K, Cl and S tends to increase with residence time, sufficient residence time is needed to ensure complete char burn-out.


34

Page 35 of 57

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

ASSOCIATED CONTENT Contents of material supplied as Supporting Information: SEM-EDS analysis of particles samples obtained at different locations in the sampling system: cyclone (Mcy), aerosol filter (Mfa), and hoses/pipes (M3), respectively; STA procedure for determination of combustible matter; Figures with elemental mass balances for K, Cl, S, P, Ca and Si; release of P as function of residence time. This material is available free of charge via the Internet at http://pubs.acs.org. AUTHOR INFORMATION Corresponding Author *Corresponding author: E-mail: [email protected], Tel: +45 4525 2800, Fax: +45 4588 2258

ACKNOWLEDGMENT The work was funded by the Danish Strategic Research Center for Power Generation from Renewable Energy (GREEN), and Biofuels Research Infrastructure for Sharing Knowledge (BRISK), and was carried out at the Combustion and harmful Emissions Control (CHEC) Research Centre, Department of Chemical and Biochemical Engineering, Technical University of Denmark. NOMENCLATURE: A0, ash content of dry fuel (used in Eq. 6)

[%]

Ai, mean ash content in the material collected in the cyclone (used in Eq. 6)

[%]


35

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 36 of 57

B, char burn-out according to Eq. (6)

[%]

c0, mass concentration of the fuel ash, or an element in the fuel ash,

e.g. [mg/g] or [g/g]

cg, mass concentration of an element in the flue gas,

e.g. [mg/g] or [g/g]

ccy, mass concentration of the ash collected in the cyclone, or an element in the ash collected in the cyclone, e.g. [mg/k] or [g/g] corg,0, fraction of organic matter in the fuel (by weight),

e.g. [g/g]

corg,cy, fraction of organic matter in the material collected in the cyclone (by weight),

e.g. [g/g]

cfa, mass concentration of the ash collected in the aerosol filter, or an element in the ash collected in the aerosol filter, e.g. [mg/g] or [g/g] M0, mass of fuel,

[g]

Mg, mass of flue gas,

[g]

Mcy, mass of particles collected in the cyclone,

[g]

Mfa: mass of particles collected in the aerosol filter,

[g]

M1, mass of particles not sampled by the probe,

[g]

M2, mass of deposit on top of sampling probe,

[g]

M3, mass of particles collected in sampling line (hoses/pipes),

[g]

M4, mass of coarse particles not collected,

[g]

M5, mass of condensed salts in probe tip,

[g]

M6, mass of not collected aerosol,

[g]

ήs, sampled fraction (fraction of gas and particles sampled by the extraction probe); ήa, fraction of sample gas directed through the aerosol filter; ήw, fraction of water-soluble K, Cl or S determined in the material collected in the cyclone; ήrecov,ash, recovered ash fraction in experiment compared to input fuel ash; ήrecov,element, recovered fraction of an element in experiment compared to input fuel; ήrel (method 1), released fraction of an element as calculated according to equation scheme (3); ήrel (method 2), released fraction of an element as calculated according to equation scheme (4);

REFERENCES


36

Page 37 of 57

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

(1) The Danish Climate Policy Plan. Towards a low carbon society. The Danish Government, 2013, ISBN 978-87-93071-29-2 (2) Zheng, Y.; Jensen, P. A.; Jensen, A. D.; Sander, B.; Junker, H. Ash transformation during co-firing of coal and straw. Fuel 2007, 86, 1008-1020 (3) Korbee, R.; Shah, K. V.; Cieplik, M.; Betrand, C. I.; Vuthaluru, H. B.; van de Kamp, W. L. First line transformations of coal and biomass during pf combustion. Energy Fuels 2010, 24, 897-909 (4) Wu, H.; Glarborg, P.; Frandsen, F. J.; Dam-Johansen, K.; Jensen, P. A.; Sander, B. Cocombustion of pulverized coal and solid recovered fuel in an entrained flow reactor – General combustion and ash behaviour. Fuel 2011, 90, 1980-1991 (5) Hansen, S. B.; Jensen, P. A.; Frandsen, F. J.; Wu, H. Deposit probe measurement in large biomass-fired grate boilers and pulverized-fuel boilers. Energy Fuels 2014, 28, 3539-3555 (6) Haynes, B. S.; Neville, M.; Quann, R. J.; Sarofim, A. F. Factors governing the surface enrichment of fly ash in volatile trace species. J. Colloid Interface Sci. 1982, 87, 266–278 (7) Helble, J. J. Mechanisms of ash particle formation and growth during pulverized coal combustion. Thesis (Ph. D.), Massachusetts Institute of Technology, Dept. of Chemical Engineering, 1987. (8) Wiinikka, H.; Gebart, R.; Boman, C.; Boström, D.; Öhman, M. Influence of fuel as composition on high temperature aerosol formation in fixed bed combustion of woody biomass pellets. Fuel 2007, 86, 181-193.


37

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 38 of 57

(9) Vejahati, F.; Xu, Z.; Gupta, R. Trace Elements in Coal: Associations with Coal and Minerals and Their Behavior during Coal Utilization-A Review. Fuel 2010, 89, 904–911 (10) Frandsen, F. J. Ash Formation, Deposition and Corrosion When Utilizing Straw for Heat and Power Production. Doctoral Thesis, Technical University of Denmark, Department of Chemical Engineering, 2010, 341 pages, ISBN: 978-87-92481-40-5 (11) Flagan, R. C.; Seinfeld, J. H. Fundamentals of Air Pollution Engineering. 1988. PrenticeHall, Inc. Englewood Cliffs, New Jersey. ISBN 0-13-332537-7 (12) Damoe, A. J.; Wu, H.; Frandsen, F. J.; Glarborg, P.; Sander, B. Impact of Coal Fly Ash Addition on Combustion Aerosols (PM2.5) from Full-Scale Suspension-Firing of Pulverized Wood. Energy Fuels 2014, 28, 3217−3223 (13) Shah, K. V.; Cieplik, M. K.; Bertrand, C. I.; van de Kamp, W. L.; Vuthaluru, H. B. Correlating the effects of ash elements and their association in the fuel matrix with the ash release during pulverized fuel combustion, Fuel Proces. Technol. 2010, 91, 531-545 (14) Knudsen, J. N.: Volatilization of inorganic matter during combustion of annual biomass, Ph.D. Dissertation, Department of Chemical Engineering, Technical University of Denmark, Lyngby, 2004. ISBN: 87-91435-11-0 (15) Knudsen, J. N.; Jensen, P. A.; Dam-Johansen, K. Transformation and release to the gas phase of Cl, K, and S during combustion of annual biomass, Energy Fuels 2004, 18, 1385-1399 (16) Knudsen, J. N.; Jensen, P. A.; Lin, W.; Dam-Johansen, K. Secondary capture of chlorine and sulfur during thermal conversion of biomass, Energy Fuels 2005, 19, 606-617


38

Page 39 of 57

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

(17) van Lith, S.C. Release of inorganic elements during wood-firing on a grate, Ph.D. Dissertation, CHEC Research Center, Department of Chemical Engineering, Technical University of Denmark, 2005. ISBN: 87-91435-29-3 (18) van Lith, S. C.; Alonso-Ramírez, V.; Jensen, P. A.; Frandsen, F. J.; Glarborg, P. Release to the gas phase of inorganic elements during wood combustion. Part 1: Development and evaluation of quantification methods, Energy Fuels 2006, 20, 964-978 (19) van Lith, S. C.; Jensen, P. A.; Frandsen, F. J.; Glarborg, P. Release to the gas phase of inorganic elements during wood combustion. Part 2: Influence of fuel compostion, Energy Fuels 2008, 22, 1598-1609 (20) Frandsen, F. J.; van Lith, S.C.; Korbee, R.; Yrjas, P.; Backman, R.; Obernberger, I.; Brunner, T.; Jöller, M. Quantification of the release of inorganic elements from biofuels. Fuel Proces. Technol. 2007, 88, 1118 – 1128 (21) Novaković, A.; van Lith, S. C.; Frandsen, F. J.; Jensen, P. A.; Holgersen, L. B. Release of potassium from the systems K-Ca-Si and K-Ca-P, Energy Fuels, 2009, 23, 3423-3428 (22) Saleh, S. B.; Flensborg, J. P.; Shoulaifar, T. K.; Sarossy, Z.; Hansen, B. B.; Egsgaard, H.; DeMartini, N.; Jensen, P. A.; Glarborg, P.; Dam-Johansen, K. Release of Chlorine and Sulfur during Biomass Torrefaction and Pyrolysis. Energy Fuels, 2014, 28, 3738−3746 (23) Jensen, P. A.; Frandsen, F. J.; Dam-Johansen, K.; Sander, B. Experimental Investigation of the Transformation and Release to Gas Phase of Potassium and Chlorine during Straw Pyrolysis, Energy Fuels 2000, 14, 1280-1285


39

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 40 of 57

(24) Hansen, S. B. Model for deposition build-up in biomass boilers, PhD thesis, Department of Chemical and Biochemical Engineering, Technical University of Denmark, 2015. (25) Hansen, S. B.; Jensen, P. A.; Frandsen, F. J.; Sander, B.; Glarborg, P. Mechanistic model for ash deposit formation in biomass suspension-firing. Part 1: Model verification by use of entrained flow reactor experiments, Energy Fuels 2016, In Press, doi: 10.1021/acs.energyfuels.6b01659 (26) Hansen, S. B.; Jensen, P. A.; Frandsen, F. J.; Sander, B.; Glarborg, P. Mechanistic model for ash deposit formation in biomass suspension-fired boilers. Part 2: Model verification by use of full scale tests, Energy Fuels 2016, In Press, doi: 10.1021/acs.energyfuels.6b01660 (27) Sarofim, A. F.; Howard, J. B.; Padia, A. S. The Physical Transformation of the Mineral Matter in Pulverized Coal under Simulated Combustion Conditions. Combust. Sci. Technol. 1977, 16, 187-204 (28) Li, B.; Sun, Z.; Li, Z.; Aldén, M.; Jakobsen, J. G.; Hansen, S.; Glarborg, P. Post-flame gas-phase sulfation of potassium chloride. Combust. Flame 2013, 160, 959-969


40

Page 41 of 57

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

TABLES. Table Title” (Word Style “VD_Table_Title”). Table 1: Fuel analysis.

Qualitative description

Straw 1

Straw 2

Wood 1

Wood 2

Danish wheat straw from 2006, pulverized

Straw pellets from BRISK project. Milled in ball mill and sieved through a 710 µm sieve

Pulverized wood pellets from Avedøre Power Plant

Softwood pellets from BRISK project. Milled in cutter mill equipped with a 750 µm sieve/knife

Water

%, as received

7.4

12.5 ± 0.6

9.7 ± 0.3

7.9 ± 0.3

Ash

%, dry basis

4.2

4.6 ± 0.2

1.0 ± 0.2

0.4 ± 0.2

Volatiles

%, dry basis

75.9

76.5 ± 1.5

83.3 ± 1.7

84.6 ± 1.7

Higher heating value

MJ/kg, basis

dry 18.92

18.97 ± 0.12

20.22 ± 0.12

20.36 ± 0.12

Lower heating value

MJ/kg, basis

dry 17.65

17.66

18.87

19.02

C

%, dry basis

46.9

48.7 ± 1.5

50.8 ± 1.5

52.2 ± 1.5

H

%, dry basis

6.0

6.0

6.2 ± 0.2

6.2

N

%, dry basis

0.56

-

0.17 ± 0.07

-

S

%, dry basis

0.12

0.080 ± 0.008

0.013 ± 0.003

0.005 ± 0.003

Cl

%, dry basis

0.65

0.18 ± 0.02

0.004 ± 0.001

0.004 ± 0.001

Al

%, dry basis

0.044

0.023 0.0020

Ca

%, dry basis

0.230

0.360 ± 0.020

Fe

%, dry basis

0.041

0.018 0.0022

K

%, dry basis

1.4

Mg

%, dry basis

0.096

± 0.0126 0.0013

± 0.0058 0.0006

0.228 ± 0.014

±

0.085 ± 0.005

± 0.0081 0.0008

± 0.0041 0.0002

±

0.8 ± 0.05

0.0926 0.0056

± 0.0410 0.002

±

0.063 ± 0.004

0.035

± 0.011 ± 0.001


41

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 42 of 57

0.0021 Na

%, dry basis

0.023

0.028 ± 0.003

0.0028 0.0005

± < 0.0011

P

%, dry basis

0.091

0.075 ± 0.006

0.0122 0.0010

± 0.0041 0.0008

Si

%, dry basis

0.39

1.10 ± 0.1

0.100 0.0100

± 0.020 ± 0.010

Ti

%, dry basis

0.0042

0.0017 0.0004

Cl/K

mol/mol

0.51

S/K

mol/mol

K/Si Ca/Si

± 0.00086 0.00038

± < 0.0005

0.25

0.05

0.11

0.10

0.12

0.17

0.15

mol/mol

2.58

0.52

0.67

1.47

mol/mol

0.41

0.23

1.60

2.98

69.1

97.9

79.1

103

244

515

298

608

662

1540

675

1710

Particle size Dv(10), µm distribution (by laser Dv(50), µm diffraction) Dv(90), µm

±


42

Page 43 of 57

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Table 2: Experimental matrix for combustion experiments in the EFR. The experimental ash balances (% recovery) are calculated as ήrecov,ash*100 % (from equation (1)) Experimental variables (settings) Fuel

Straw 1

Straw 2

Wood 1

Wood 2

Ash balances at end of experiment

Reactor temp.

Fuel feed rate

Gas % ash in residence cyclone time material

Ash recovery, experimental

(°C)

(g/h)

(s)

(%)

(%)

Straw 1 1200 2s

1200

295

2

88.0

45

Straw 1 1200 1.1s

1200

295

1.1

79.0

80

Straw 1 1200 0.28s

1200

295

0.28

27.0

85

Straw 1 1400 1.7s

1400

295

1.7

90.0

43

Straw 1 1400 1s

1400

295

1

73.0

49

Straw 1 1400 0.25s

1400

295

0.25

32.0

75

Straw 2 1400 1.7s

1400

295

1.7

98

67

Straw 2 1400 1s

1400

295

1

95

78

Wood 1 1200 2s

1200

267

2

70.0

48

Wood 1 1200 1.1s

1200

267

1.1

38.5

57

Wood 1 1400 1.7s

1400

267

1.7

81.0

62

Wood 1 1400 1s

1400

267

1

42.0

48

Wood 1 1400 0.25s

1400

267

0.25

3.5

98

Wood 2 1400 1.7s

1400

267

1.7

30

62

Wood 2 1400 1s

1400

267

1

2.5

153

Preheater N2 quench temp. flow (L/min): (°C):

Main air flow: (L/min):

Feeder air flow: (L/min):

Purge air flow: (L/min):

1100

15

10

5

Exp. short name

Fixed settings All experiments

5.48


43

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 44 of 57

Table 3. Water soluble fractions, ήw,, of K, Cl and S determined in cyclone ash for each fuel (average of several measurements). We believe that at least at higher residence times (≥ 1s) the water-soluble species found in the cyclone ash are probably from recondensation of gas phase salt species during sample cooling; and thus the water-soluble fraction provides a reasonable estimate of the released species.

Water soluble K fraction in cyclone ash (ήw) Cl S

Straw 1

Straw 2

Wood 1

Wood 2

0.8

0.4

0.7

0.8

0.99

0.95

0.99

(0.95)1)

0.95

0.99

0.90

0.90

Italic indicates that the release data for the elements have not been plotted, as the uncertainty on the results is considered too high. 1) Estimated value (analytical data uncertain, as some values are below detection limit. Please refer to Figure 3 for details on total and water soluble concentrations of inorganic elements).

FIGURES (Word Style “VA_Figure_Caption”).


44

Page 45 of 57

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Figure 1: The DTU solid fuel entrained flow reactor (left) and the gas/particle sampling system (right).


45

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Not sampled fraction Gas + Particles M1 = (1-ήs)∙M0

Deposit on probe top, M2

Coarse particles >1.5 µm

Particles collected in hoses/pipes, M3

Fuel c0M0 Combustion Sampled fraction Gas + Particles ήs

Not collected material, M4

Residual ash

Gas (HCl, SO2 etc.) cgMg

Fine aerosol particles

Fly Ash Formation during Suspension Firing of Biomass: Effects of

Recommend Documents