Chemical study of fly ash deposition in combustion of pelletized residual agricultural biomass

Chemical study of fly ash deposition in combustion of 1 pelletized residual agricultural biomass 2 Javier Royo; Paula Canalís; David Quintana; 3 University of Zaragoza, c/ María de Luna 3, E-50018 Zaragoza, Spain 4 5 ABSTRACT 6 Agricultural residual biomass has great potential as an energy source, but is used only to a limited 7 extent mainly because of the characteristics of its ash (quantity and composition), which can lead to 8 problematic phenomena during combustion, among them fly ash deposition, the focus of this study. 9 A previous work presented the results of laboratory experiments carried out using a fixed-grate 10 reactor and involving four different agropellets under different operating conditions; the variables 11 tested were deposition rate, bottom ash proportion and sintering degree during combustion. Based 12 on these results, the analysis has been taken further and the fly ash deposits collected during these 13 tests have been characterized by SEM-EDS and XRD. A methodology to differentiate between 14 deposits caused by condensation (including thermophoresis and turbulent diffusion) and by inertial 15 impact of coarse fly ash entrained from the bed has been proposed. Deposition by condensation has 16 been found to decrease for higher values of excess air ratio in all cases. Conversely, deposition by 17 inertial impact does not show a common behavior, due to the influence of bottom ash sintering 18 degree and fuel composition. The ultimate aim of this study is to gain a better understanding of fly 19 ash deposition, in order to develop better fuel blends, boiler design and operating parameters, 20 enhancing the market penetration of agricultural residual biomass. 21 22 KEY WORDS: 23 Agricultural residual biomass; Combustion; Fixed bed reactor; Fly ash deposition; SEM-EDS; XRD 24 25 1. INTRODUCTION 26 The main contribution of biomass to the generation of renewable energy in the EU is found in the 27 heating and cooling sector [1], where important growth is expected in coming years; the target for 28 2020 having been set at 3785 PJ [1]. New uses for forest biomass ([2], [3]), in addition to the 29 traditional energy production, make imperative to find new resources with which to meet the 30 predictable rise in demand for thermal energy. The biggest growth in supply should come from the 31 agricultural sector, where an increase of over 150% compared with 2006 is expected [1]. 32 In addition to energy crops and some types of residual agro-industrial biomass, these new resources 33 mainly comprise agricultural crop residues: herbaceous crop residues and pruning residues of 34 permanent woody crops. In particular, this paper focuses on three residual agricultural biomasses: 35 vineyard pruning residues, corn stover and barley straw. These were selected due to their potential 36 as sources of energy both in Europe and the rest of the world. FAOSTAT data (available at [4]) 37 indicate that the area covered by vines and maize and barley crops in the EU in 2017 was nearly 23.4 38 Mha. Using conservative availability indices (50% for vineyard pruning residue and corn stover, and 39 10% for barley straw), this translates into an energy potential of over 500 PJ/yr for the EU. In 40 consequence, their use could contribute significantly to achieve the objectives set. 41 Thermal conversion of agricultural biomass, mainly of the herbaceous type, shows clear differences 42 compared with forest biomass. This is mainly due to the characteristics of the ash (quantity and 43 composition), which can lead to certain problems in conversion facilities. 44 During combustion, ash undergoes physical and chemical transformations which cause fractioning. 45 Part of the components of the ash remain as a solid fraction which accumulates in the grate (bottom 46 ash) and in some cases can sinter, affecting conversion in the bed, restricting efficiency of the grate 47 and negatively affecting the control of gaseous emissions: carbon monoxide, nitrogen oxides, and 48 volatile organic compounds ([5]-[9]). 49 Other part, mainly related with alkali metal compounds, is volatilized. After complex and not always 50 well known mechanisms ([10], [11]), these compounds can directly condense or after forming 51 aerosols be deposited by thermophoresis and/or turbulent diffusion ([11]-[13]) on the surfaces of the 52 equipment used for heat exchange, in the form of small crystals (e.g., potassium chloride -KCl-, 53 potassium sulfate -K2SO4and potassium carbonates -K2CO3 and KHCO3-). An ash entrainment of solid 54 particles (coarse fly ash) in gas combustion flow from the bed can also be generated and, in some 55 conditions, these particles can be deposited on convective areas by inertial impact. These 56 phenomena (volatilization and ash entrainment) are responsible, alongside deposition, for corrosion 57 and erosion, which reduce equipment performance and use-life ([5], [9]). 58 In recent decades, several prestigious research centers have been working towards identifying key 59 factors in the conversion of biofuels, as well as in the transformation of their ash, in order to 60 understand problems caused by the latter ([8], [14]-[23]). In all cases, the critical influence of ash 61 chemical composition, especially the concentration of Na, Mg, Al, Si, P, S, Cl, K and Ca [24], is 62 recognized in issues associated with thermal conversion (e.g. sintering, deposition, corrosion, erosion 63 and emissions). However, chemical composition is not the only factor, since ash behavior is also 64 affected by combustion conditions in the bed, which are themselves related to design ([23]-[26]) and 65 operational parameters [27]. 66 Owing to the complexity of the phenomena that contribute to ash fractioning, combustion tests are 67 often undertaken in laboratory reactors, most of which operate with a fixed-grate in order to keep 68 combustion conditions under control ([23],[28]-[32]). This type of reactor enables, in the simplest 69 way, the collection of important information concerning the behavior of fuels under different 70 operating conditions. It allows the evaluation of fuel reactivity (ignition front velocity and ignition 71 rate [33]), quantifying of bottom ash in the bed and determining its propensity to sintering, as well as 72 quantifying the amount of solid residue deposited on heat exchange surfaces per time and unit area 73 (deposition rate) [31]. Furthermore, these reactors allow samples to be taken for the 74 characterization of solid residues (bottom ash fraction and fly ash deposits), allowing a better 75 understanding of the phenomena driving ash fractioning. In addition, the analysis of gaseous (e.g. 76 CO, NOx or volatile organic compounds) ([34]-[37]) and particle emissions ([38]-[40]), is also possible. 77 In a previous work [41], authors presented the results analysis of the four first points (reactivity, 78 bottom ash quantity, sintering degree and deposition rate) for different pellets made of residual 79 agricultural biomass (agropellets). These pellets were evaluated under a range of operating 80 conditions in a laboratory fixed-grate reactor. In this paper, it is intended to go a step further and 81 characterize fly ash deposition samples collected in combustion tests by means of scanning electron 82 microscopy (SEM) with energy dispersive X-ray spectrometry (EDS), and powder X-ray diffractometry 83 (XRD). These methods are widely used to identify and characterize ash compounds ([9],[13], [42]-84 [48]). 85 SEM-EDS provides detailed imaging information about morphology, as well as defining the elemental 86 chemical composition of samples. This technique is both easy and highly precise. Although elements 87 which are present in concentrations below 0.1-0.5% are below detection limits [42], in general it 88 does not affect the detection of the previously commented most significant ash-forming elements 89 responsible for ash-related operational problems during combustion. 90 The XRD method is applied to identify and quantify crystalline phases present in the sample by 91 measuring their concentrations, as well as determining the amorphous fraction [42]. 92 These are complementary techniques. On the one hand, XRD allows for a better understanding of 93 how chemical elements detected by SEM-EDS are associated. On the other hand, the identification of 94 minor minerals in a multicomponent system by means of XRD is uncertain due to such issues as 95 detection limits, peak overlapping and unknown amorphous matter. SEM-EDS results facilitate the 96 identification of phases and can provide confirmation of XRD results [49]. 97 From the results obtained by means of SEM-EDS and XRD, a methodology is proposed to 98 differentiate between deposits caused by condensation (including thermophoresis and turbulent 99 diffusion) and by inertial impact of coarse fly ash entrained from the bed. 100 The ultimate aim is to gain a better understanding of deposition phenomena affecting agricultural 101 residual biomass. This will, it is hoped, help researchers and technologists to make better decisions 102 regarding fuel blends, boilers design and optimum operating parameters, increasing the market 103 penetration of this important type of biomass. 104 2. MATERIAL AND METHODS 105 2.1 Fuels 106 Fly ash deposition chemistry of four different agropellets (agricultural residual pellets) is studied in 107 this paper: 108 • Woody agropellet: 100% Vineyard pruning pellet (PV)1 109 • Mixed agropellets (Vineyard pruning blended with an herbaceous component): 110 o 70% Vineyard pruning + 30% Barley straw (PVB) 111 o 70% Vineyard pruning + 30% Corn stover (PVC) 112 o 60% Vineyard pruning + 20% Corn stover + 20% Barley straw (PVCB). 113 The main thermochemical properties of selected fuels are reproduced from [41] and shown in Tables 114 1 and 2. 115 Table 1 116 Fuel properties (% m/m: mass percentage; d.b.: dry basis; w.b.: wet basis). 117 PV PVB PVC PVCB Bulk density (kg·m-3)a 599 562 556 546 Proximate analysis (% m/m d.b.) Volatile matter b 76.5 72.4 72.1 72.3 Fixed carbonc 20.5 21.7 18.6 21.2 Ashd 3.1 5.9 9.3 6.5 Total moisture (% m/m w.b.)e 9.0 9.1 9.2 9.0 Ultimate analysis (% m/m d.b.) Carbonf 48.9 46.36 46.01 46.36 Hydrogen f 5.8 5.77 5.64 5.55 Nitrogen f 0.55 0.56 0.55 0.60 Sulfur g 0.09 0.055 0.050 0.094 Chlorine g 0.03 0.047 0.080 0.090 Oxygen c 41.6 41.29 38.33 40.58 HHV (d.b. at p=constant) (MJ·kg-1)h 19.11 18.54 18.06 18.36 LHV (w.b. at p=constant) (MJ·kg-1)h 16.01 15.48 15.06 15.40 a EN 15103:2009 b EN-ISO 18123:2016 c Calculated d EN-ISO 18122:2016 e EN-ISO 18134:2016 118 f EN-ISO 16948:2015 g EN-ISO 16994:2015 h EN-ISO 14918:2011 119 1 Vineyard pruning residues used to produce this agropellet were not the same as those used for mixed agropellets. 120 Table 2 121 Ash properties (% m/m: mass percentage; d.b.: dry basis). 122 PV PVB PVC PVCB Chemical ash composition (% m/m d.b.)a Al2O3 0.91 2.72 2.19 2.30 CaO 42.39 45.77 48.17 40.54 Fe 2 O 3 0.71 2.22 1.98 1.27 K2O 30.09 14.88 15.79 19.43 MgO 10.45 8.64 7.64 11.01 Na2O 0.62 0.41 0.39 0.38 P 2 O 5 7.35 4.45 4.00 4.36 SO3 3.95 2.32 3.24 4.39 SiO2 2.65 17.70 15.31 15.22 TiO 2 0.07 0.17 0.18 0.16 Cl 0.12 0.21 0.57 0.54 Ash melting points in oxidizing conditions (oC)b Initial deformation temperature (DT) 1240 1130 1310 1330 Hemisphere temperature (HT) > 1500 1310 1460 1460 Flow temperature (FT) > 1500 1370 1480 1470 a EN-ISO 16967:2015 b CEN/TS 15370-1:2006 123 2.2 Reactor 124 As noted, research of ash-related phenomena during combustion is generally carried out with the aid 125 of laboratory reactors. In the case of fixed-bed reactors, simplified geometries are used to being able 126 to consider one-dimensional behavior [50]. 127 In order to perform the combustion tests, an experimental fixed-grate reactor was used (see Fig. 1). 128 In this reactor, inlet air is injected through the grate from the bottom by means of a fan equipped 129 with a variable-frequency drive which allows airflow to be regulated. Since experiments require inlet 130 air temperature to remain under control, the reactor is equipped with a refrigerator and an electrical 131 resistor either to cool the air or heat it as needed. This allowed two different types of tests to be 132 undertaken: without preheating (inlet air at 25oC) and with preheating (inlet air at 80oC). The reactor 133 is fitted with fifteen N thermocouples to monitor temperature both at the bed and the freeboard. 134 In addition, the facility includes a deposition probe, with a removable sampling ring in the chimney of 135 the reactor [51]. This is a common device used to simulate fly ash deposition in furnace pipes and 136 heat exchangers [49]. Prior to the experiment, the removable sampling ring is cleaned, dried, 137 measured and weighed. During the stable combustion period, the deposition probe is inserted inside 138 the chimney and the ring is cooled by compressed air, keeping its surface at an appropriate 139 temperature for studying deposition [51]. For the tests presented here, compressed inlet air was 140 adjusted to keep an average temperature of 335±25oC. Once extracted, the dirty ring is dried and 141 weighed again to determine the mass of deposits, allowing deposition rate (DR, g·m-2·h-1) to be 142 calculated and, thus, the different propensity of each fuel used for deposition to be assessed ([31], 143 [41], [52]-[56]). 144 145 Fig. 1. Scheme of the experimental test facility [41]. 146 Finally, once combustion is completed and the reactor cools down, bottom ash is collected from the 147 surface of the grate for weighing and classification, which allows the sintering tendency of each fuel to 148 be determined ([5], [21], [41], [57], [58]). Three fractions were considered: S1, which passes through a 149 3.15 mm sieve and is considered to be not sintered; S2, which does not pass through a 3.15 mm sieve, 150 but is easily disaggregated by hand and presents a low sintering degree; S3, which does not pass the 151 3.15 mm sieve, is difficult to disaggregate by hand and presents a high sintering degree. Since the 152 difference between S2 and S3 is subjective, a fraction S2/3 encompassing both classes was used. 153 2.3 Ash analysis 154 In all the tests, once deposits had been weighed and deposition rate calculated, a sample was taken 155 from the front face of the removable sampling rings, that is, from the side facing and perpendicular 156 to the flow of combustion gases. Samples of S1 bottom ash fractions were also collected. All samples 157 were glued onto metal plates with carbon tape and coated with carbon before being analyzed by 158 SEM-EDS. The equipment used was a Carl Zeiss Merlin electronic field emission microscope equipped 159 with Gemini Column, with acceleration voltages between 0.02 and 30 kV, fitted with an EDS X-MAS 160 detector by Oxford Instruments with a window of 20 mm2 and energy resolution between 127 eV 161 and 5.9 keV. For each sample, three 1 mm2-zones were selected, and images taken with the retro-162 dispersed detector (asb). Average elemental composition was obtained through EDS, using a voltage 163 of 15 kV. INCA software was used to process the results. Major participating elements in the most 164 important ash transformation processes -namely Na, Mg, Al, Si, P, S, Cl, K, Ca and Fewere included 165 in the analysis. 166 In addition, four combustion experiments without air preheating were selected for each fuel. These 167 tests were chosen to cover evenly the common range of excess air ratio (λ) for each fuel (see Table 168 3). A preheated experiment was also selected for each agropellet, all four with an almost identical 169 excess air ratio value (λ≈1.3). For all these tests the crystalline matter composition of the fly ash 170 deposition samples collected in the ring was determined by XRD. Standard X-ray diffraction patterns 171 were collected at room temperature using a Rigaku D/max instrument with a copper rotating anode 172 and a graphite monochromator to select CuKα wavelength. The measurements were performed at 173 40 kV and 80 mA, in the angular range from 5° to 80° on 2θ, applying a step size of 0.03° and a 174 counting rate of 1 s/step. X-ray patterns were analyzed with JADE software, with access to the JCPDS-175 International Centre for Diffraction Database (2000) and profile-based RIR analyses. 176 3. RESULTS AND ANALYSIS 177 3.1. Tests and results 178 A total of 68 combustion tests were carried out following the same protocol with the four fuels. As 179 already noted tests both with and without preheating (“ph” experiments) – varying inlet air 180 temperature (Ta) – were undertaken for every fuel. In the tests the excess air ratio ranged from 1.1 181 to 2.3 (over-stoichiometric conditions), in order to reproduce the combustion conditions found in 182 small domestic equipment. Table 3 summarizes the main features of the experiments performed. 183 Table 3 184 Outline of test features. 185 PV PVB PVC PVCB Number of tests performed Without preheating (Ta=25 °C) 10 10 10 12 Preheated tests (-ph-; Ta=80 °C) 8 6 6 6 λ Min 1.15 1.21 1.18 1.23 Max 2.04 2.30 2.29 2.07 Fed fuel (kg) 4.03 3.78 3.74 3.67 186 Table 4 shows mean values (and range) of elemental composition obtained by SEM-EDS for each of 187 the four fuels of fly ash deposits, expressed as a percentage of the total mass of measured elements 188 (Na, Mg, Al, Si, P, S, Cl, K, Ca and Fe). 189 Furthermore, Fig. 2 plots the values summarized in Table 4 against excess air ratio, keeping out the 190 elements with a concentration significantly lower than 10 % in all the samples (Al, Si, P and Fe) of fly 191 ash deposits collected in the combustion experiments (with and without preheating). Due to their 192 chemical similarity and the almost identical role they play in the reactions that take place in ash 193 transformation processes, the concentrations of K and Na [24] as well as Ca and Mg [59] have been 194 aggregated in Fig. 2. 195 The alternate combination of deposits by condensation and by inertial impact results in the 269 construction of an overlapping multi-layered structure ([12], [60]). 270 In the following subsection, it is proposed a methodology that allows estimating the percentage and 271 the amount of ash deposited by each of these two processes. 272 3.3.1. Methodology description 273 The two mechanisms involved in ash deposition are highly complex, and it is helpful to stablish 274 several simplifications in order to facilitate the analysis of results. 275 A first set of assumptions is related to bottom ash and the entrainment of coarse fly ash. It is 276 assumed that Si forms compounds (silicates, aluminosilicates and oxides) that remain solid regardless 277 of combustion temperatures ([12], [24]). Consequently, all Si present in the sampling ring (subscript 278 “Probe”) is assumed to have been deposited by inertial impact (subscript “Imp.”)2: 279 SiProbe = SiImp. (1) 280 It has been reported that the chemical composition of coarse fly ash entrained from the bed 281 resembles that of bottom ash ([52], [64]). For the development of this methodology it is only 282 necessary to consider that Si/K proportion in S1 fraction (subscript “S1”, a fraction that is constituted 283 by particles that can be easily dragged) remains the same in entrained ash (which can subsequently 284 be deposited by inertial impact). This assumption, together with equation (1), leads to the following 285 equation (2): 286 (K/Si)S1 = (K/Si)Imp. = KImp. /SiProbe (2) 287 Following equation (2), and considering that, out of the total amount of K found in sampling ring 288 deposits, a fraction can be ascribed to inertial impact of solids entrained directly from the bed, with 289 the remainder ascribed to condensation (subscript “Cond.”), equation (3) follows: 290 KCond. = KProbe - KImp. = KProbe – (K/Si)S1 · (Si)Probe (3) 291 The argumentation that leads to equations (2) and (3) would also work for Na instead of K. 292 2 In equations (1) to (3) “Si” and “K” can be mass or molar contents of each element. A second set of considerations is related to the compounds present in deposits. Based on XRD results 293 (see Table 5), crystalline phases related chlorides, sulfates and carbonates of alkali metals 294 (condensation) are KCl, K2SO4, KHCO3, while those related to Mg, Si and Ca (inertial impact) are 295 CaCO3, Ca(OH)2, MgO, SiO2 and Ca2SiO4. 296 Bearing in mind the molecular mass of the aforementioned compounds, it can be noted that: 297 • 1 kg of KCond. present in the probe implies 1.9 kg of KCl, 2.23 kg of K2SO4 or 2.56 kg of 298 KHCO3. That is to say, 1 kg of KCond. implies deposits of the order of 2 kg forming KCl, 299 K2SO4 and/or KHCO3. Other chlorides, sulfates and carbonates of alkali metals which can 300 be volatilized and then deposited by condensation, even those which do not appear in 301 these XRD results (e.g. K2CO3 and K3Na(SO4)2), also follow, in order of magnitude, the 302 proportion of 2 kg of deposits per kg of K+Na3. 303 • 1 kg of Ca present in the probe implies 1.85 kg of Ca(OH)2 or 2.5 kg of CaCO3; 1 kg of Mg 304 implies 1.67 kg of MgO; 1 kg of Si implies 2.14 kg of SiO2; 1 kg of Ca+Si (0.74 kg of Ca + 305 0.26 kg of Si) implies 1.59 kg of Ca2SiO4. That is to say, 1 kg of Ca+Mg+Si also implies 306 deposits of the order of 2 kg in the form of MgO, CaCO3, Ca(OH)2, SiO2 and/or Ca2SiO4. 307 Likewise, the compounds of Ca, Mg, Si, P, Al and/or Fe, which are typically formed in 308 combustion and can be entrained from the bed (mainly silicates, aluminosilicates, 309 phosphates, oxides, carbonates, sulfates, and hydroxides, where K and Na can also be 310 present [65]), also follow, in order of magnitude, the proportion of 2 kg of deposits per 311 kg of Ca+Mg+Si+P+Al+Fe+K+Na4. 312 In view of the fact that, on one side, each mass unit of KCond.+NaCond. and, on the other side, of 313 Ca+Mg+Si+P+Al+Fe+ KImp.+NaImp. produces approximately (in terms of order of magnitude) the same 314 amount of deposits, it is possible to approximately determine the mass ratio of deposits due to 315 3 If the 7 compounds of this type mentioned in [65] with a presence over 1 % are considered, it is obtained a maximum value of 2.52, minimum of 1.77 and mean (unweighted) of 2.19 kg of deposits per each kg of K+Na (deposited by condensation). 4 If the 45 compounds of this type mentioned in [65] with a presence over 1 % are considered, it is obtained a maximum value of 3.40, minimum of 1.40 and mean (unweighted) of 1.93 kg of deposits per each kg of Ca+Mg+Si+P+Al+Fe+K+Na (for alkalis, the fraction from deposits by inertial impact is only included). condensation (DMCond., kg) and inertial impact (DMImp, kg) according to equation (4)5: 316 DMCond./DMImp.= (K+Na)Cond./((Ca+Mg+Si+P+Al+Fe)Probe + (K+Na)Imp.) (4) 317 Taking into account the relationship shown in equation (4), the mass fraction of deposits caused by 318 condensation (mf_depositsCond.) and inertial impact (mf_depositsImp.) can be accounted for, as shown 319 in equations (5) and (6): 320 mf_deposits Cond. = DMCond./(DMCond. + DMImp.) = 321 = (K+Na)Cond./((K+Na)Cond. + ((Ca+Mg+Si+P+Al+Fe)Probe + (K+Na)Imp.)) = 322 = (K+Na)Cond./(K+Na+Ca+Mg+Si+P+Al+Fe)Probe (5) 323 324 mf_depositsImp. = DMImp./(DMCond.+ DMImp.) = 325 = ((Ca+Mg+Si+P+Al+Fe)Probe + (K+Na)Imp.)/( (K+Na)Cond. + ((Ca+Mg+Si+P+Al+Fe)Probe + (K+Na)Imp.)) = 326 = ((Ca+Mg+Si+P+Al+Fe)Probe + (K+Na)Imp.)/(K+Na+Ca+Mg+Si+P+Al+Fe)Probe (6) 327 By multiplying each of these mass fractions by the deposition rate it is possible to share out the total 328 mass of deposits between both mechanisms, obtaining the deposition rate by condensation (DRCond., 329 g·m-2·h-1) and by inertial impact (DRImp., g·m-2·h-1), equations (7) and (8): 330 DRCond. = mf_depositsCond. · DR (7) 331 DRImp. = mf_depositsImp. · DR (8) 332 Naturally, this methodology only provides approximate values, but based on some reasonable 333 hypotheses and simplifications, it allows obtaining conclusions about the mechanisms of deposition 334 (condensation or inertial impact), as can be seen in next subsection. 335 3.3.2. Methodology application and discussion 336 The combination of the elemental composition (SEM-EDS) of ring deposits (Fig. 2 and Table 4) and S1 337 bottom ash fractions (Fig. 3), the deposition rate [41] and equations (3) to (8), leads to Fig. 4, which 338 expresses deposition rates by condensation and by inertial impact as a function of excess air ratio for 339 each of the four fuels analyzed (tests with and without preheating are shown). 340 5 In equations (4) to (6) “Ca”, “Mg”, “Si”, “P”, “Al”, “Fe”, “K” and “Na”, are mass contents of each element. 341 Fig. 4. Deposition rates (DR) by condensation and by inertial impact against excess air ratio (λ) to PV, PVB, PVC and PVCB.342 Regarding tests without inlet air preheating, it can be verified that, for all fuels analyzed, deposition 343 by condensation clearly decreases as excess air ratio increases (for all fuels and within the range of λ 344 analyzed, there is a factor of about 3 between the highest and the lowest condensation deposition 345 rate values). The reason for this lies in the fact that a greater excess air ratio leads to a reduction of 346 combustion temperature, limiting the volatility of the main reactive ash elements that play a role in 347 deposition by condensation (mainly K, Cl and S) ([66], [67]). 348 In addition, the substantial values of deposition by condensation presented by PVB, especially at low 349 excess air ratios, are noteworthy. This is confirmed by the chemical analysis of the deposits: 350 • XRD analysis (see Table 5): PVB has a high percentage of KCl (37.4% on average in the four 351 samples analyzed, much higher than the other pellets) and K2SO4 (30.1% on average in the 352 four samples analyzed, only below PVCB). 353 • SEM-EDS analysis (see Fig.2 and Table 4): PVB presents the highest percentages of K and Cl, 354 and also of S. 355 However, there is no obvious correlation between these results and ash properties obtained in fuel 356 analysis (see Tables 1 and 2), as PVB presents lower concentrations of K and Cl than the other mixed 357 pellets, and lower concentration of S than PVCB. This fact corroborates that trying to predict the 358 performance of ash biomass based only on indices obtained from fuel analysis is not always accurate; 359 in fact, the usefulness of these indices has been questioned in other research works (e.g., [48], [56], 360 [68]-[70]). 361 Continuing with the results of tests without preheating, they also reveal that PV presents slightly 362 lower values of deposition rate by condensation than mixed pellets (≤ 13 g·m-2·h-1 in all cases), 363 especially when excess air ratios are low. This fact is related to the different composition of these 364 deposits, which present low percentages of K2SO4 and, above all, of KCl detected by XRD (Table 5), 365 although this is partially compensated by the high percentage of KHCO3 (25.3%, whereas in the rest 366 of fuels it is only detected in one PVB sample). The low concentration of KCl in the deposits can be 367 explained by the low concentration of Cl in this fuel (Tables 1 and 2). Concerning K2SO4, although PV 368 presents a high percentage of K and S (Tables 1 and 2), K has a greater affinity for P (it is found in very 369 significant amounts in this fuel (Table 2)), what could facilitate the formation of K-phosphates before 370 K-sulfates [24]. 371 Concerning deposition by inertial impact, no common tendency has been found to apply to all fuels 372 in tests without preheating. Whereas in PV its value clearly decreases when excess air ratio is higher, 373 in mixed pellets it remains practically constant. These different tendencies among fuels may be 374 caused by the fact that an increase in excess air ratio leads to two opposite effects that interact with 375 different weights: on the one hand, increasing the air flow raises its speed in the bed, encouraging 376 the entrainment of coarse fly ash; on the other hand, the adhesion of solid particles is discouraged, 377 as sticky deposits in the ring (alkali metal sulfates and chlorides) become less substantial, owing to 378 reduced vaporization and subsequent condensation. 379 To deepen the analysis of the behavior presented by each fuel, Table 6 shows total deposition rate, 380 by condensation and by inertial impact, together with bottom ash proportion and sintering degree 381 (fraction S2/3) as reflected by the experimental results presented in [41]. 382 Table 6 383 Bottom ash proportion, sintering degree and deposition rates (total, by condensation and by inertial impact) 384 mean values (range) of all tests without inlet air preheating. 385 Bottom ash proportion a Sintering degree (fraction S2/3) a DR DRCond. DRImp. % % g·m-2·h-1 g·m-2·h-1 % g·m-2·h-1 % PV 25.3 (18.1-31.3) 1.6 (0.3-3.1) 16.6 (11.0-21.3) 8.1 (3.7-13.0) 49.0 (31.0-62.5) 8.5 (5.5-12.8) 51.0 (37.5-69.0) PVB 74.7 (72.8-77.7) 51.8 (26.7-62.2) 20.9 (10.5-29.9) 17.3 (8.2-24.3) 82.9 (73.4-90.8) 3.6 (1.9-5.6) 17.1 (9.2-26.6) PVC 50.0 (48.3-50.7) 33.8 (23.5-40.7) 19.4 (14.2-25.0) 10.1 (5.4-15.5) 52.0 (32.9-74.1) 9.3 (5.3-14.0) 48.0 (25.4-67.1) PVCB 59.5 (58.4-60.9) 40.1 (26.4-49.8) 19.2 (13.5-23.8) 9.7 (4.3-16.1) 54.0 (30.1-68.8) 8.3 (6.2-10.8) 46.0 (31.2-69.9) a % with regard to total mass of ash introduced with the fuel. 386 Table 6 seems to indicate that, in mixed pellets, sintering prevents deposition by inertial impact by 387 discouraging the entrainment of particles from the bed. As a result, PVB, which presents high 388 sintering values, yields a much lower deposition rate by inertial impact values than PVC and PVCB. 389 The results of inlet air preheating tests (represented by unfilled markers in Fig. 4) can be used to 390 corroborate this behavior of mixed pellets. Inlet air preheating increases air velocity in the bed, 391 encouraging entrainment, but does not lead to a significant increase of combustion temperature 392 [41]: 393 • Due to the fact that combustion temperature remains practically unchanged, ash 394 vaporization and therefore deposition by condensation is not affected substantially by 395 preheating for any of the mixed pellets. 396 • In contrast, it may be observed that air preheating largely increases deposition by inertial 397 impact in PVC and PVCB, but much less in the case of PVB. In other words, an increase in air 398 velocity causes a much bigger impact in particle entrainment and its subsequent deposition 399 in fuels which are less susceptible to sintering. 400 The case of PV is somewhat different, because high entrainment (there is very little ash retention in 401 the bed) does not directly translate, in tests without preheating, into a greater amount of deposits by 402 inertial impact compared to that of other fuels, possibly owing to the lower quantity of sticky 403 deposits. In fact, it seems that given that this fuel presents low concentrations of KCl and K2SO4, the 404 sampling ring was saturated by deposits from inertial impact. As a result of this saturation, following 405 an increase of excess air ratio and hence a decrease in sticky deposits due to condensation, the 406 capacity of PV to retain deposits by inertial impact decreases (Fig. 4). It is worth analyzing the results 407 of the tests with preheating for this fuel. First, a sharp decrease in deposition by condensation can be 408 noticed compared with tests without preheating. This is due to the lower combustion temperatures 409 reached in experiments with preheating [41], which discourages the evaporation of alkali metals 410 from the bed, as expected in view of Table 5. The fact that KOH has greater affinity for SO2/SO3 and 411 HCl than for CO2 [24] explains at least partly the aforementioned non-appearance of K-carbonates in 412 the preheated test. In addition, preheating of inlet air entails no significant increase of deposition by 413 inertial impact, which reinforces the idea of saturation. 414 415 3.3.3. (K+Na)/(Cl+2S) molar ratios 416 To complete the comparative analysis of the various fuels, Table 7 shows (K+Na)/(Cl+2S) molar ratios, 417 calculated from SEM results for deposit samples (Fig. 2), equation (3) and the initial analysis of the 418 fuels (Tables 1 and 2). 419 Table 7 420 (K+Na)/(Cl+2S) molar ratios (mean values of tests with and without preheating). 421 (K+Na)Probe/(Cl+2S) deposits SEM (K+Na)Cond./(Cl+2S) deposits SEM (K+Na)/(Cl+2S) Fuel analysis PV 2.13 1.71 3.16 PVB 1.26 1.21 4.08 PVC 1.11 1.01 6.02 PVCB 1.24 1.09 3.29 422 Regarding (K+Na)Probe/(Cl+2S) molar ratio using total alkali metal concentration in deposits by SEM, it 423 can be observed that all fuels present values higher than 1, i.e., there is an excess of alkali metals 424 compared with Cl and S. The reasons for this are twofold: 425 • Part of K and Na have vaporized as hydroxides and condensed as carbonates (mainly KHCO3, 426 which was detected by XRD in PV and in one PVB sample, see Table 4). 427 • Ash entrainment of solid particles containing K and Na from the bed occurs. 428 If the second ratio is considered, (K+Na)Cond./(Cl+2S) -after discounting alkali metals compounds 429 deposited in the sampling ring by inertial impact (following equation (3))- it can be noted that, in the 430 case of PVCB and PVC, the value obtained is very close to 1 (practically all alkali metal have 431 condensed as chlorides or sulfates), while in PVB and especially in PV it is higher, owing to the 432 aforementioned presence of KHCO3. The coherence of these ratios with XRD results corroborates 433 that the hypotheses and assumptions on which equation (3) was based were sound. 434 Finally, it should be stressed that it is not possible to easily predict the values of these molar ratios 435 (obtained by analyzing the deposits) nor to explain differences in the behavior of the various fuels on 436 the basis of ratios calculated with fuels preliminary analysis (last column of Table 7), since there is no 437 direct correspondence between them. 438 4. CONCLUSIONS 439 This study has presented the results of chemical analysis of deposits obtained in combustion tests 440 carried out with four varieties of agropellet in a laboratory fixed-grate reactor. The analyses were 441 carried by electron microscopy (SEM) with energy dispersive X-ray spectrometry (EDS) and X-ray 442 diffractometry (XRD). 443 In order to take the results further, a simple methodology was developed that allows for deposits 444 produced by condensation (including thermophoresis and turbulent diffusion) and by inertial impact 445 of coarse fly ash entrained from the bed to be distinguished. 446 This methodology, alongside the results of chemical analysis and the data for deposition rates 447 presented in a previous work [41], has yielded important results concerning the deposition 448 phenomena affecting the four agropellets under study. 449 It was confirmed that an increase in excess air leads to a decrease in deposition by condensation, 450 owing to a reduction in combustion temperatures, which limits the volatility of K, Cl and S. The lower 451 deposition rates attested for PV could be related to its high P content. 452 However, concerning deposition by inertial impact, no common behavior has been found, probably 453 because an increase in excess air ratio leads to two opposite effects. First, an increase in excess air 454 ratio also increases the air flow, encouraging the entrainment of coarse fly ash. This effect becomes 455 less acute as sintering increases; although sintering undermines the operation of the grate, it also 456 discourages ash entrainment. Second, an increase in air excess ratio leads to a decrease of deposits 457 by condensation, some of which take the shape of a sticky layer (mainly alkali metal sulfates and 458 chlorides), and thus the adhesion of coarse fly ash entrained from the bed. In fact, at least 459 concerning PV, it is argued that the adhesion of solid particles to sticky deposits can result in 460 saturation. 461 The quantification of deposits produced by condensation and by inertial impact, although achieved 462 through a series of simplifications and assumptions, provides useful information which, it is hoped, 463 will contribute to finding solutions to the problem posed by high deposition rates in the combustion 464 of agricultural residual biomass, leading to both better fuel blends and boiler design and operational 465 parameters, increasing the market penetration of this important kind of biomass. 466 As noted, fly ash deposition and bottom ash sintering are related; sintering and the relationship 467 between both phenomena will be addressed in depth from bottom ash chemical characterization, as 468 part of complementary further research studies. 469 5. ACKNOWLEDGEMENTS 470 The authors greatly acknowledge the Spanish Ministry of Science, Innovation and Universities for 471 funding the project “MHWPellet: Mixed pellets based on agricultural crops residues (herbaceous and 472 woody) for their use in the residential sector: optimization of their composition and conversion 473 parameters” (ref. ENE2015-68809-R (MIMECO/FEDER, UE)). 474 Authors also would like to acknowledge the use of Servicio General de Apoyo a la Investigación-SAI, 475 Universidad de Zaragoza. 476 6. REFERENCES 477 [1] Scarlat N, Dallemand JF, Monforti-Ferrario F, Banja M. Renewable energy policy framework 478 and bioenergy contribution in the European Union – An overview from National Renewable 479 Energy Action Plans and Progress Reports. Renewable and Sustainable Energy Reviews 2015; 480 51: 969-985. 481 [2] Scarlat N, Dallemand JF, Monforti-Ferrario F, Nita V. The role of biomass and bioenergy in a 482 future bioeconomy: Policies and facts. Environmental Development 2015; 15: 3-34. 483 [3] Bozell JJ, Black SK, Myers M, Cahill D, Miler WP, Park S. Solvent fractionation fo renewable 484 woody feedstocks: Organoslv generation of biorefinery process streams for the production of 485 biobased chemicals. Biomass Bioenergy 2011; 35(10): 4197-208 486 [4] Food and Agricultural Organization of the United Nations – FAOSTAT - 487 http://www.fao.org/faostat/ , accessed in May 2019 488 [64] Díaz-Ramírez MC. Grate-Fired Energy Crop Conversion: Experiences with Brassica Carinata and 638 Populus sp. Ed. Springer Theses 2015; ISBN 978-3-319-20758-2 639 [65] Vassilev S.V., Baxter B., Vassileva C.G. An overview of the behaviour of biomass during 640 combustion: Part I. Phase-mineral transformations of organic and inorganic matter. Fuel, 2013; 641 112:391–449 642 [66] Theis M., Skrifvars B.-J., Zevenhoven M., Hupa M., Tran H. Fouling tendency of ash resulting 643 from burning mixtures of biofuels. Part 2: deposit chemistry. Fuel, 2006; 85: 1992–2001. 644 [67] Thy P, Jenkins BM, Grundvig S, Shiraki R, Lesher CE. High temperature elemental losses and 645 mineralogical changes in common biomass ashes. Fuel, 2006; 85:783–95. 646 [68] Vassilev S. V., Baxter D., Vassileva C. G. An overview of the behaviour of biomass during 647 combustion: Part II. Ash fusion and ash formation mechanisms of biomass types. Fuel, 2014; 648 117: 152–183 649 [69] García-Maraver A., Zamorano M., Fernandes U., Rabaçal M,. Costa M. Relationship between 650 fuel quality and gaseous and particulate matter emissions in a domestic pellet-fired boiler. 651 Fuel, 2014; 119: 141-152 652 [70] Xiao R., Chen X., Wang F., Yu G. The physicochemical properties of different biomass ashes at 653 different ashing temperature. Renewable Energy, 2011; 36: 244-249 654