- Email: [email protected]
Limestone fragmentation and attrition during fluidized bed oxyfiring
Fuel 89 (2010) 827–832
Contents lists available at ScienceDirect
Fuel journal homepage: www.elsevier.com/locate/fuel
Limestone fragmentation and attrition during fluidized bed oxyfiring Fabrizio Scala a, Piero Salatino b,* a b
Istituto di Ricerche sulla Combustione – CNR, Piazzale Tecchio 80, 80125 Napoli, Italy Dipartimento di Ingegneria Chimica, Università degli Studi di Napoli Federico II, Piazzale Tecchio 80, 80125 Napoli, Italy
a r t i c l e
i n f o
Article history: Received 12 September 2008 Received in revised form 19 February 2009 Accepted 19 March 2009 Available online 9 April 2009 Keywords: Fluidized bed combustion Oxyfiring Attrition Fragmentation Desulfurization
a b s t r a c t Attrition/fragmentation of limestone under simulated fluidized bed oxyfiring conditions was investigated by means of an experimental protocol that had been previously developed for characterization of attrition/fragmentation of sorbents in air-blown atmospheric fluidized bed combustors. The protocol was based on the use of different and mutually complementary techniques. The extent and pattern of attrition by surface wear in the dense phase of a fluidized bed were assessed in experiments carried out with a bench scale fluidized bed combustor under simulated oxyfiring conditions. Sorbent samples generated during simulated oxyfiring tests were further characterized from the standpoint of fragmentation upon high velocity impact by means of a purposely designed particle impactor. Results showed that under calcination-hindered conditions attrition and fragmentation patterns are much different from those occurring under air-blown atmospheric combustion conditions. Noteworthy, attrition/fragmentation enhanced particle sulfation by continuously regenerating the exposed particle surface. Ó 2009 Elsevier Ltd. All rights reserved.
1. Introduction Attrition and fragmentation of limestone during the fluidized bed (FB) combustion of sulfur-bearing fuels have been thoroughly characterized over the last decade [1–7]. Key phenomenological features and mechanistic pathways of sorbent attrition/fragmentation in air-blown atmospheric FB combustors have been disclosed with the aid of a comprehensive test protocol consisting of different and mutually complementary test procedures [1,3,7]. Primary fragmentation occurs as a consequence of thermal stresses and internal overpressures due to carbon dioxide emission. Attrition by abrasion and secondary fragmentation are related to mechanical stresses due to rubbing and collisions with other particles or with the reactor walls or internals. Particle breakage upon impact provides another pathway to sorbent fragmentation, particularly in the jetting region of the bottom bed and in the cyclone inlet ducts. The progress of chemical reactions (calcination, sulfation) interferes with the attrition/fragmentation processes, making the phenomenology even more complicated. The mechanisms and extent of attrition/fragmentation of sorbents under oxyfiring conditions have only recently received consideration. Under these conditions, which are gaining much interest in connection to carbon capture and sequestration (CCS) techniques, carbon dioxide partial pressures established in the reaction zone may be large to the point of preventing limestone calcination. It has been shown [8] that primary fragmentation is less pronounced under calcination-hindered conditions. Sulfur * Corresponding author. Tel.: +39 081 7682258; fax: +39 081 5936936. E-mail address: [email protected] (P. Salatino). 0016-2361/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.fuel.2009.03.024
capture is also significantly affected, the main pathway being direct sulfation of the raw limestone. Patterns and extent of sorbent attrition and impact fragmentation are also likely to be affected. Shimizu et al. [5] reported that under pressurized FB combustion conditions (where large CO2 partial pressures establish) sulfur capture was controlled by the limestone attrition rate. In fact, since most of the limestone in the boiler captures sulfur dioxide at a very slow rate under product layer SO2 diffusion control, attrition of the sorbent particles surface may reduce the thickness of the calcium sulfate layer and, in turn, increase the reaction rate. The present study addresses attrition and fragmentation of limestone under simulated FB oxyfiring conditions. Experiments were carried out with a bench scale FB combustor to investigate the extent and pattern of attrition by surface wear in the dense phase of a fluidized bed and the influence of parallel limestone sulfation on this phenomenon. Sorbent samples generated during simulated oxyfiring tests have been further characterized as regards their fragmentation behavior under high velocity impact conditions in a purposely designed particle impactor. In a forthcoming paper [9], the phenomenology of attrition/fragmentation observed under oxyfiring conditions will be compared with that recorded during air-blown atmospheric FB combustion.
2. Experimental 2.1. Apparatus Sulfation of limestone was carried out in a stainless steel atmospheric bubbling FB reactor. The reactor, 40 mm ID and 1 m high,
828
F. Scala, P. Salatino / Fuel 89 (2010) 827–832
was electrically heated. The gas distributor was a perforated plate with 55 holes of 0.5 mm diameter in a triangular pitch. CO2, O2 and a SO2–N2 mixture were separately supplied to the reactor by means of digital mass flow meters/controllers. Flue gases were continuously sampled at the exit for CO2 and SO2 concentration measurement by means of two on-line NDIR analyzers, in order to monitor the progress of reactions. A two-exit head was purposely designed to convey flue gases through either of two 25 mm ID cylindrical sintered brass filters (filtration efficiency = 1 for >10 lm-particles). Alternated use of filters enabled time-resolved capture of elutriated fines at the exhaust. The bed material consisted of mixtures of 20 g of limestone and either 150 g of silica sand or 200 g of corundum (corresponding to an unexpanded bed height of 8.3 cm). Silica sand was sieved in the two particle size ranges 0.2–0.3 mm and 0.85–1.0 mm, corundum in the particle size range 0.2–0.3 mm. Minimum fluidizing velocity at 850 °C was 0.02 and 0.28 m s1 for the two sand size ranges, respectively, and 0.03 m s1 for corundum. Further details on the apparatus can be found elsewhere [1]. Impact testing of pre-processed sorbent samples was carried out in the test apparatus represented in Fig. 1. Particles are entrained in a gas stream at controlled velocity and impacted against a target [7]. The test rig consists of a vertical stainless steel eductor tube (1 m high, 10 mm ID) equipped with a particle feeder. The particle feeding device is fitted at the top of the eductor tube and consists of a stainless steel hopper with a 10 mm ID at the top section and 4 mm ID at the bottom section. The bottom end of the hopper is connected through a valve to a steel tube (6 mm OD and 4 mm ID) running coaxially for 0.2 m inside the eductor tube. Air enters the top section of the eductor tube and flows downwards between the inner and outer tubes. When the valve is opened, limestone particles contained in the hopper flow through the inner tube driven by gravity and by the draw induced by air flowing in the eductor tube. The hopper can be isolated from the environment by means of a top valve, to avoid air bypass when
LIMESTONE INLET
2 3 4 5
AIR OUTLET
6 1
9
8
the bottom valve is open. After feeding, the particles are accelerated by the air flow in the eductor tube. The particle velocity is controlled by regulating the air flow in the eductor tube, by means of a flowmeter. When the particles exit the eductor tube, they impact on a rigid target plate placed in a collection chamber 50 mm below the tube exit. The target is made of stainless steel and is inclined by 30° with respect to the horizontal. This inclination was chosen as a trade off between the need of avoiding interference between the impacting and reflected particles and the need of minimizing the departure of results from those obtained with a target perpendicular to the particle trajectory [10]. The collection chamber is made of a glass vessel, 0.55 m high and 80 mm ID. The air flow leaves the collection chamber from the top section where it passes through a porous cellulose filter (for the capture of finer particulate). The impacted limestone particles settle at the bottom. The device is designed so as to minimize the loss of limestone entrained by air in the chamber, enabling easy collection of the limestone particles after impact for further analysis. The particle impact velocity was calculated as the sum of the gas velocity in the eductor tube and the particle terminal velocity. Particle acceleration to this velocity is complete before impact, as confirmed by particle tracking at the exit of the eductor tube with a high-speed (10,000 frames per second) video camera Photron Ultima APX. 2.2. Procedures A high-calcium (96.8%) Italian limestone (Massicci) was used in the experiments. Fresh limestone particles were sieved in the particle size range 0.4–0.6 mm, falling well within the range of particle sizes that are customarily employed in practical circulating FB combustion. Limestone particles were then sulfated (at 2000 ppmv SO2) to completion in the FB reactor operated batchwise at 850 °C with a gas superficial velocity of either 0.51 or 0.73 m s1. In all the tests CO2 inlet concentration was kept at 71% by volume. In this environment, at atmospheric conditions, limestone calcination is prevented as verified by XRD analysis of the exhausted sorbent, which did not show any presence of CaO. At the end of each test the sorbent was first cooled down in a high-CO2 environment and then discharged from the bed. The sorbent was easily sieved out of the bed material because of its smaller or larger particle size. Sulfated limestone particles were again sieved in the particle size range 0.4–0.6 mm and stored in a desiccator to prevent hydration of the material. Samples (approximately 2.0 g) of sulfated limestone were weighed and used for fragmentation tests in the impact testing apparatus. The tests were carried out in air at ambient temperature with the following particle impact velocities v: 10, 17, 24, 31, 38 and 45 m s1. These velocities were selected so as to reproduce impact conditions that are likely to establish near the gas distributor or in the cyclone inlet ducts of industrial-scale circulating FB combustors. After each test the sample was retrieved from the collection chamber and weighed. Closure of the mass balance was checked to estimate the loss of material during testing. The closure was always within 3% of the initial sample weight. The collected particles were then sieve-analyzed to obtain their particle size distribution. Selected tests were repeated to check the reproducibility of the particle size distributions.
10 3. Results
AIR
7
Fig. 1. Impact damage test apparatus. (1) gas flowmeter; (2–4) lock hopper valves; (3) hopper; (5) feeding tube; (6) eductor tube; (7) target plate; (8) collection chamber; (9) cellulose filter and (10) gas flow metering valve.
3.1. Calcium conversion degree Fig. 2 reports the overall degree of calcium conversion XCa as a function of time during sulfation tests carried out in beds of either
829
F. Scala, P. Salatino / Fuel 89 (2010) 827–832
3.2. Elutriation rate
0.30
Inert bed: 0.2 - 0.3mm Sand 0.85 - 1.0mm Sand 0.2 - 0.3mm Corundum
0.25
X Ca, -
0.20
0.15 0.15
0.12 0.10
0.11
0.05
0.00 0
50
100
150
200
250
Time, min Fig. 2. Degree of calcium conversion as a function of time during FB limestone sulfation tests at 850 °C with different inert beds at 0.73 m s1.
sand or corundum fluidized at 0.73 m s1. XCa is defined as the cumulative (molar) SO2 uptake divided by the initial (molar) calcium inventory. According to this definition, XCa embodies contributions from sulfation of coarse particles as well as of attrited fines. Operating conditions were such that the reactor could not be considered differential with respect to SO2. In fact, SO2 concentration at the exhaust dropped to vanishingly small values at the beginning of the test, to increase thereafter until the inlet concentration was approached in the long term. This feature prevented the assessment of the kinetics of sulfur capture from differential analysis of XCa vs. t curves. The degree of calcium conversion of the sorbent after 240 min was 11–12% with the bed of sand (depending on the sand particle size) and 15% with corundum. These values are significantly smaller than those found for the same limestone under conventional atmospheric FB combustion conditions (about 30% [1]). This is consistent with other results reported in the literature which show that direct sulfation of limestone yields lower conversion degrees in the short term than the sulfation of calcined limestone [11]. It is noteworthy that corundum promotes sulfation of limestone as compared with sand. This is most likely due to more extensive abrasion promoted by corundum which contributes to disclose unconverted calcium for further sulfation.
Fig. 3 reports the sorbent elutriation rate E(t) measured during sulfation at the two fluidization velocities tested. The elutriation rate was normalized with respect to the initial amount of limestone fed to the reactor (W0 = 20 g). A pronounced spike is recorded at the beginning of the test, related to the initial removal of surface asperities (rounding off) from the angular coarse sorbent particles. As rounding off is complete, E(t) drops to a minimum at 5–15 min whence it steadily increases along with the progress of sulfation (see Fig. 2). Notably, the increase of E(t) with the progress of sulfation is opposite to what observed under atmospheric FB combustion conditions, where sulfation hinders attrition and elutriation by strengthening the particle outer layer [1,3]. This difference can be explained by assuming that the sulfur-rich shell formed during direct sulfation of limestone is somewhat softer and more prone to attrition than the original calcium carbonate. Comparison of data reported in Fig. 3A and B indicates that E(t) increases with the gas superficial velocity and decreases with the inert particle size, both features consistent with previous results obtained under different operating conditions [1]. The elutriation rate with corundum is larger than that with sand, consistently with conversion data reported in Fig. 2. It must be underlined, however, that E(t) is only partly related to actual sorbent attrition rate, as part of the attrited material is likely to have been retained in the reactor during the experiments. Elutriation data have been analyzed according to the following equation:
E¼k
0.00030
Inert bed:
A
Inert bed:
B
Sand (0.85-1.0mm) Corundum (0.2-0.3mm)
0.00025 0.00020 0.00015 0.00010
Sand (0.85-1.0mm) Sand (0.2-0.3mm) Corundum (0.2-0.3mm)
0.00025
E/W 0, min -1
E/W 0, min -1
ð1Þ
which relates sorbent elutriation rate to the sorbent loading W, its average particle size d and the gas superficial velocity U. The characteristic velocity U0 in Eq. (1) represents a threshold above which bubble-induced attrition occurs, typically larger than the incipient fluidization velocity Umf of the bed material [12]. Data points in Fig 3A and B have been worked out to yield the instantaneous value of the attrition rate constant k, defined according to Eq. (1) with the approximation W ffi W 0 and d ffi d0 . Values of U0 giving the best fit to the actual data were 0.4 and 0.35 m/s for bed materials in the size ranges 0.3–0.4 mm and 0.85–1.0 mm, respectively. It is remarkable that these values are significantly larger than Umf, a feature that might be related to the high mechanical strength of the sorbent throughout conversion. Values of k obtained with this procedure
0.00030
0.00020 0.00015 0.00010 0.00005
0.00005 0.00000
W U U0 d
0
20
40
60
80
100
Time, min
120
140
160
180
0.00000
0
20
40
60
80
100
120
140
160
180
Time, min
Fig. 3. Limestone elutriation rate as a function of time during FB sulfation tests at 850 °C with different bed materials at gas superficial velocities of 0.51 m s1 (A) and 0.73 m s1 (B).
830
F. Scala, P. Salatino / Fuel 89 (2010) 827–832
1e-8
1e-8
Inert bed: Sand
A
B
U = 0.51m s-1 (0.85-1.0mm)
8e-9
Inert bed: Corundum U = 0.51m s-1
8e-9
U = 0.73m s-1 (0.85-1.0mm) U = 0.73m s-1 (0.2-0.3mm)
6e-9
U = 0.73m s-1
k, -
k, -
6e-9
4e-9
4e-9
2e-9
2e-9
0
0
20
40
60
80
100
120
140
160
0
180
0
20
40
60
Time, min
80
100
120
140
160
180
Time, min
Fig. 4. Sorbent attrition constant k, defined according to Eq. (1), as a function of time during FB sulfation tests at 850 °C at different gas superficial velocities. Bed material: (A) sand and (B) corundum.
are plotted as a function of time in Fig. 4A and B for beds of sand and corundum, respectively. 3.3. Impact fragmentation Samples to be subjected to impact tests were obtained by FB sulfation for 240 min. Sorbent particles were sieved again after sulfation and only the fraction remaining within the 0.4–0.6 mm size interval was subject to impact testing. Figs. 5A and 6A show the probability density function (PDF) of the fragments size for sulfated particles subjected to impact fragmentation tests at different impact velocities for particles pre-processed in beds of sand and corundum, respectively. The same distributions are presented as cumulative fractional mass in Figs. 5B and 6B. By ‘‘fragments” we mean all the particles collected after the impact tests which size falls below the lower limit of the feed size interval: accordingly all the collected particles finer than 0.4 mm were classified as ‘‘fragments” in the context of the present study. Since Figs. 5B and 6B only report the cumulative distributions of the fragments the curves do not end at 1.0, but at a value representing the fractional mass of fragments (the complement to 1.0 being the fractional mass of particles in the size range 0.4–0.6 mm). The PDF was obtained by dividing the fractional mass of particles in a given size bin by the width (in mm) of the size bin.
The extent of fragmentation increases as the impact velocity increases from 10 to 45 m s1. Figs. 5B and 6B show that the fractional amount of fragments is smaller than 10% and 30% of the initial sample mass for sorbent sulfated in sand and corundum, respectively, even at the largest impact velocity of 45 m s1. The extent of impact fragmentation of sorbents pre-processed in a bed of corundum (Fig. 6B) is much larger than that observed for the sand bed (Fig. 5B). Moreover, the effect of increasing v on the extent of particle fragmentation is more pronounced in the former case. Figs. 5A and 6A better highlight the way fragment sizes are distributed. At any impact velocity a dominance of relatively coarse fragments, of size just slightly smaller than 0.4 mm, is observed for the bed of sand (Fig. 5A). On the other hand, a bimodal fragment size distribution is evident as the sorbent is processed in a bed of corundum, with a significant fraction of particles below 100 lm. This difference in the PDF curves is indicative of different fragmentation patterns of the sulfated limestone in the two cases. Fig. 7 summarizes the results obtained with the different samples. Data are reported as the total fractional mass f of fragments (i.e., the total mass of fragments finer than 0.4 mm collected after the impact test) versus impact velocity. The log–log plot was chosen so as to better highlight the establishment of power-law f vs. v relationships (f / vk) that are often used to correlate these data [13]. Inspection of data in Fig. 7 suggests the following features:
2.0
0.30
A
-1
pdf, mm
-1
1.5
-1
cumulative size distribution, -
v=10m s v=17m s-1 v=24m s-1 -1 v=31m s -1 v=38m s -1 v=45m s
1.0
0.5
0.0
B
v=10m s v=17m s-1 -1 v=24m s -1 v=31m s v=38m s-1 v=45m s-1
0.25
0.20
0.15
0.10
0.05
0.00 0.0
0.1
0.2
0.3
particle diameter, mm
0.4
0.5
0.0
0.1
0.2
0.3
0.4
0.5
particle diameter, mm
Fig. 5. Probability density function of particle size (A) and cumulative size distribution (B) of fragments (d < 0.4 mm) collected after impact at different velocities of batches of limestone particles sulfated at 850 °C and 0.73 m s1 with an inert bed of 0.85–1.0 mm sand.
831
F. Scala, P. Salatino / Fuel 89 (2010) 827–832
2.0
0.30
A
-1
cumulative size distribution, -
pdf, mm
-1
1.5
v=10m s v=17m s-1 v=24m s-1 -1 v=31m s -1 v=38m s -1 v=45m s
1.0
0.5
0.0 0.0
v=10m s-1 -1 v=17m s v=24m s-1 -1 v=31m s -1 v=38m s v=45m s-1
0.25
0.20
B
0.15
0.10
0.05
0.00 0.1
0.2
0.3
0.4
0.5
0.0
particle diameter, mm
0.1
0.2
0.3
0.4
0.5
particle diameter, mm
Fig. 6. Probability density function of particle size (A) and cumulative size distribution (B) of fragments (d < 0.4 mm) collected after impact at different velocities of batches of limestone particles sulfated at 850 °C and 0.73 m s1 with an inert bed of 0.2–0.3 mm corundum.
the effect of pre-processing on fragmentation is much emphasized as the impact velocity is raised above 20 m s1; curves relative to the same bed material (corundum vs. sand) tend to overlap at high impact velocity, suggesting that the nature of the bed material plays a much larger role as compared with bed solids particle size and/or gas superficial velocity.
1 -1
0.85 - 1.0mm Sand, 0.51m s
-1
0.2 - 0.3mm Sand, 0.73m s
-1
0.85 - 1.0mm Sand, 0.73m s
0.2 - 0.3mm Corundum, 0.51m s-1 0.2 - 0.3mm Corundum, 0.73m s-1
f, -
0.1
The last point is further supported by data in Fig. 8. Here, the Sauter mean diameters of sorbent particles remaining after impact at different impact velocities decrease to a much larger extent for samples pre-processed in a bed of corundum than in a bed of sand, displaying a minor influence of bed solids size and gas superficial velocity.
0.01
0.001 1
10
100
v, ms
-1
Fig. 7. Fraction of fragmented material as a function of the impact velocity for limestone particles sulfated at 850 °C and subjected to impact fragmentation tests.
Sauter mean diameter, µ m
600
500
400
300
200
0.85 - 1.0mm Sand, 0.51m s-1 0.2 - 0.3mm Sand, 0.73m s-1 0.85 - 1.0mm Sand, 0.73m s-1 0.2 - 0.3mm Corundum, 0.51m s-1 0.2 - 0.3mm Corundum, 0.73m s-1
100
0 0
10
20
30
40
50
-1
v, ms
Fig. 8. Sauter mean diameter of limestone particles sulfated at 850 °C after impact as a function of the impact velocity.
at low impact velocity the extent of fragmentation, as measured by the fragmentation index f, is rather limited, and barely influenced by the pre-processing conditions;
4. Conclusions The propensity of limestone to undergo attrition by surface wear and fragmentation by impact loading under oxyfiring fluidized bed combustion conditions has been assessed by means of a combination of test protocols. Attrition by abrasion is extensive at the very beginning of sorbent sulfation in a fluidized bed, due to initial rounding off of angular raw limestone particles, to decay immediately thereafter. Due to the larger attrition propensity of partly sulfated limestone compared with the raw unconverted one, the progress of sulfation brings about an increase of particle attrition rate. Attrition rate increases with increasing gas superficial velocity and size of bed inert material. The comparison of results of experiments in which limestone was sulfated in a bed of corundum with those in which sulfation occurred in a bed of sand highlight the potential of intentionally promoting attrition to enhance sulfation. Particle sulfation, relatively fast at first, slows down as sulfation proceeds. In the long term, the progress of sulfation appears to be closely linked to the parallel progress of attrition, which makes unconverted calcium accessible for further sulfation. Using corundum as bed material significantly increases sulfur uptake over the same sulfation time. Attrition by impact of pre-processed limestone particles is strongly dependent on particle impact velocity. Limestone samples prepared in a bed of corundum are characterized by a much larger propensity to undergo impact damage when compared with those pre-processed in a bed of sand. This feature is reflected both by the extent and particle size distribution of the debris generated upon each impact.
832
F. Scala, P. Salatino / Fuel 89 (2010) 827–832
Altogether, the phenomenology and mechanisms of limestone attrition and fragmentation under oxyfiring combustion conditions differ significantly from those observed under air-blown atmospheric fluidized bed combustion conditions. Factors that promote attrition may significantly enhance the rate and extent of limestone sulfation. Acknowledgement The experimental support of A. Parlato and G. Somma is gratefully acknowledged. References [1] Scala F, Cammarota A, Chirone R, Salatino P. Comminution of limestone during batch fluidized-bed calcination and sulfation. AIChE J 1997;43:363–73. [2] Di Benedetto A, Salatino P. Modelling attrition of limestone during calcination and sulfation in a fluidized bed reactor. Powder Technol 1998;95:119–28. [3] Scala F, Salatino P, Boerefijn R, Ghadiri M. Attrition of sorbents during fluidized bed calcination and sulphation. Powder Technol 2000;107:153–67. [4] Anthony EJ, Granatstein DL. Sulfation phenomena in fluidized bed combustion systems. Prog Energ Combust Sci 2001;27:215–36.
[5] Shimizu T, Peglow M, Sakuno S, Misawa N, Suzuki N, Ueda H, et al. Effect of attrition on SO2 capture by limestone under pressurized fluidized bed combustion conditions-comparison between a mathematical model of SO2 capture by single limestone particle under attrition condition and SO2 capture in a large-scale PFBC. Chem Eng Sci 2001;56:6719–28. [6] Chen Z, Lim CJ, Grace JR. Study of limestone particle impact attrition. Chem Eng Sci 2007;62:867–77. [7] Scala F, Montagnaro F, Salatino P. Attrition of limestone by impact loading in fluidized beds. Energ Fuel 2007;21:2566–72. [8] Saastamoinen J, Pikkarainen T, Tourunen A, Räsänen M, Jäntti T. Model of fragmentation of limestone particles during thermal shock and calcination in fluidised beds. Powder Technol 2008;187:244–51. [9] Scala F, Salatino P. Flue gas desulfurization under simulated oxyfiring fluidized bed combustion conditions: the influence of limestone attrition and fragmentation. Chem Eng Sci, submitted for publication. [10] Salman AD, Biggs CA, Fua J, Angyal I, Szabó M, Hounslow MJ. An experimental investigation of particle fragmentation using single particle impact studies. Powder Technol 2002;128:36–46. [11] Hu G, Dam-Johansen K, Wedel S, Hansen JP. Review of the direct sulfation reaction of limestone. Prog Energ Combust Sci 2006;32:386–407. [12] Werther J, Reppenhagen J. Attrition. In: Yang W-C, editor. Handbook of fluidization and fluid-particle systems. New York: Marcel Dekker; 2003. p. 239–55. [13] Papadopoulos DG, Ghadiri M. Impact breakage of poly-methylmethacrylate (PMMA) extrudates. 1. Chipping mechanism. Adv Powder Technol 1996;7:183–97.
Contents lists available at ScienceDirect
Fuel journal homepage: www.elsevier.com/locate/fuel
Limestone fragmentation and attrition during fluidized bed oxyfiring Fabrizio Scala a, Piero Salatino b,* a b
Istituto di Ricerche sulla Combustione – CNR, Piazzale Tecchio 80, 80125 Napoli, Italy Dipartimento di Ingegneria Chimica, Università degli Studi di Napoli Federico II, Piazzale Tecchio 80, 80125 Napoli, Italy
a r t i c l e
i n f o
Article history: Received 12 September 2008 Received in revised form 19 February 2009 Accepted 19 March 2009 Available online 9 April 2009 Keywords: Fluidized bed combustion Oxyfiring Attrition Fragmentation Desulfurization
a b s t r a c t Attrition/fragmentation of limestone under simulated fluidized bed oxyfiring conditions was investigated by means of an experimental protocol that had been previously developed for characterization of attrition/fragmentation of sorbents in air-blown atmospheric fluidized bed combustors. The protocol was based on the use of different and mutually complementary techniques. The extent and pattern of attrition by surface wear in the dense phase of a fluidized bed were assessed in experiments carried out with a bench scale fluidized bed combustor under simulated oxyfiring conditions. Sorbent samples generated during simulated oxyfiring tests were further characterized from the standpoint of fragmentation upon high velocity impact by means of a purposely designed particle impactor. Results showed that under calcination-hindered conditions attrition and fragmentation patterns are much different from those occurring under air-blown atmospheric combustion conditions. Noteworthy, attrition/fragmentation enhanced particle sulfation by continuously regenerating the exposed particle surface. Ó 2009 Elsevier Ltd. All rights reserved.
1. Introduction Attrition and fragmentation of limestone during the fluidized bed (FB) combustion of sulfur-bearing fuels have been thoroughly characterized over the last decade [1–7]. Key phenomenological features and mechanistic pathways of sorbent attrition/fragmentation in air-blown atmospheric FB combustors have been disclosed with the aid of a comprehensive test protocol consisting of different and mutually complementary test procedures [1,3,7]. Primary fragmentation occurs as a consequence of thermal stresses and internal overpressures due to carbon dioxide emission. Attrition by abrasion and secondary fragmentation are related to mechanical stresses due to rubbing and collisions with other particles or with the reactor walls or internals. Particle breakage upon impact provides another pathway to sorbent fragmentation, particularly in the jetting region of the bottom bed and in the cyclone inlet ducts. The progress of chemical reactions (calcination, sulfation) interferes with the attrition/fragmentation processes, making the phenomenology even more complicated. The mechanisms and extent of attrition/fragmentation of sorbents under oxyfiring conditions have only recently received consideration. Under these conditions, which are gaining much interest in connection to carbon capture and sequestration (CCS) techniques, carbon dioxide partial pressures established in the reaction zone may be large to the point of preventing limestone calcination. It has been shown [8] that primary fragmentation is less pronounced under calcination-hindered conditions. Sulfur * Corresponding author. Tel.: +39 081 7682258; fax: +39 081 5936936. E-mail address: [email protected] (P. Salatino). 0016-2361/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.fuel.2009.03.024
capture is also significantly affected, the main pathway being direct sulfation of the raw limestone. Patterns and extent of sorbent attrition and impact fragmentation are also likely to be affected. Shimizu et al. [5] reported that under pressurized FB combustion conditions (where large CO2 partial pressures establish) sulfur capture was controlled by the limestone attrition rate. In fact, since most of the limestone in the boiler captures sulfur dioxide at a very slow rate under product layer SO2 diffusion control, attrition of the sorbent particles surface may reduce the thickness of the calcium sulfate layer and, in turn, increase the reaction rate. The present study addresses attrition and fragmentation of limestone under simulated FB oxyfiring conditions. Experiments were carried out with a bench scale FB combustor to investigate the extent and pattern of attrition by surface wear in the dense phase of a fluidized bed and the influence of parallel limestone sulfation on this phenomenon. Sorbent samples generated during simulated oxyfiring tests have been further characterized as regards their fragmentation behavior under high velocity impact conditions in a purposely designed particle impactor. In a forthcoming paper [9], the phenomenology of attrition/fragmentation observed under oxyfiring conditions will be compared with that recorded during air-blown atmospheric FB combustion.
2. Experimental 2.1. Apparatus Sulfation of limestone was carried out in a stainless steel atmospheric bubbling FB reactor. The reactor, 40 mm ID and 1 m high,
828
F. Scala, P. Salatino / Fuel 89 (2010) 827–832
was electrically heated. The gas distributor was a perforated plate with 55 holes of 0.5 mm diameter in a triangular pitch. CO2, O2 and a SO2–N2 mixture were separately supplied to the reactor by means of digital mass flow meters/controllers. Flue gases were continuously sampled at the exit for CO2 and SO2 concentration measurement by means of two on-line NDIR analyzers, in order to monitor the progress of reactions. A two-exit head was purposely designed to convey flue gases through either of two 25 mm ID cylindrical sintered brass filters (filtration efficiency = 1 for >10 lm-particles). Alternated use of filters enabled time-resolved capture of elutriated fines at the exhaust. The bed material consisted of mixtures of 20 g of limestone and either 150 g of silica sand or 200 g of corundum (corresponding to an unexpanded bed height of 8.3 cm). Silica sand was sieved in the two particle size ranges 0.2–0.3 mm and 0.85–1.0 mm, corundum in the particle size range 0.2–0.3 mm. Minimum fluidizing velocity at 850 °C was 0.02 and 0.28 m s1 for the two sand size ranges, respectively, and 0.03 m s1 for corundum. Further details on the apparatus can be found elsewhere [1]. Impact testing of pre-processed sorbent samples was carried out in the test apparatus represented in Fig. 1. Particles are entrained in a gas stream at controlled velocity and impacted against a target [7]. The test rig consists of a vertical stainless steel eductor tube (1 m high, 10 mm ID) equipped with a particle feeder. The particle feeding device is fitted at the top of the eductor tube and consists of a stainless steel hopper with a 10 mm ID at the top section and 4 mm ID at the bottom section. The bottom end of the hopper is connected through a valve to a steel tube (6 mm OD and 4 mm ID) running coaxially for 0.2 m inside the eductor tube. Air enters the top section of the eductor tube and flows downwards between the inner and outer tubes. When the valve is opened, limestone particles contained in the hopper flow through the inner tube driven by gravity and by the draw induced by air flowing in the eductor tube. The hopper can be isolated from the environment by means of a top valve, to avoid air bypass when
LIMESTONE INLET
2 3 4 5
AIR OUTLET
6 1
9
8
the bottom valve is open. After feeding, the particles are accelerated by the air flow in the eductor tube. The particle velocity is controlled by regulating the air flow in the eductor tube, by means of a flowmeter. When the particles exit the eductor tube, they impact on a rigid target plate placed in a collection chamber 50 mm below the tube exit. The target is made of stainless steel and is inclined by 30° with respect to the horizontal. This inclination was chosen as a trade off between the need of avoiding interference between the impacting and reflected particles and the need of minimizing the departure of results from those obtained with a target perpendicular to the particle trajectory [10]. The collection chamber is made of a glass vessel, 0.55 m high and 80 mm ID. The air flow leaves the collection chamber from the top section where it passes through a porous cellulose filter (for the capture of finer particulate). The impacted limestone particles settle at the bottom. The device is designed so as to minimize the loss of limestone entrained by air in the chamber, enabling easy collection of the limestone particles after impact for further analysis. The particle impact velocity was calculated as the sum of the gas velocity in the eductor tube and the particle terminal velocity. Particle acceleration to this velocity is complete before impact, as confirmed by particle tracking at the exit of the eductor tube with a high-speed (10,000 frames per second) video camera Photron Ultima APX. 2.2. Procedures A high-calcium (96.8%) Italian limestone (Massicci) was used in the experiments. Fresh limestone particles were sieved in the particle size range 0.4–0.6 mm, falling well within the range of particle sizes that are customarily employed in practical circulating FB combustion. Limestone particles were then sulfated (at 2000 ppmv SO2) to completion in the FB reactor operated batchwise at 850 °C with a gas superficial velocity of either 0.51 or 0.73 m s1. In all the tests CO2 inlet concentration was kept at 71% by volume. In this environment, at atmospheric conditions, limestone calcination is prevented as verified by XRD analysis of the exhausted sorbent, which did not show any presence of CaO. At the end of each test the sorbent was first cooled down in a high-CO2 environment and then discharged from the bed. The sorbent was easily sieved out of the bed material because of its smaller or larger particle size. Sulfated limestone particles were again sieved in the particle size range 0.4–0.6 mm and stored in a desiccator to prevent hydration of the material. Samples (approximately 2.0 g) of sulfated limestone were weighed and used for fragmentation tests in the impact testing apparatus. The tests were carried out in air at ambient temperature with the following particle impact velocities v: 10, 17, 24, 31, 38 and 45 m s1. These velocities were selected so as to reproduce impact conditions that are likely to establish near the gas distributor or in the cyclone inlet ducts of industrial-scale circulating FB combustors. After each test the sample was retrieved from the collection chamber and weighed. Closure of the mass balance was checked to estimate the loss of material during testing. The closure was always within 3% of the initial sample weight. The collected particles were then sieve-analyzed to obtain their particle size distribution. Selected tests were repeated to check the reproducibility of the particle size distributions.
10 3. Results
AIR
7
Fig. 1. Impact damage test apparatus. (1) gas flowmeter; (2–4) lock hopper valves; (3) hopper; (5) feeding tube; (6) eductor tube; (7) target plate; (8) collection chamber; (9) cellulose filter and (10) gas flow metering valve.
3.1. Calcium conversion degree Fig. 2 reports the overall degree of calcium conversion XCa as a function of time during sulfation tests carried out in beds of either
829
F. Scala, P. Salatino / Fuel 89 (2010) 827–832
3.2. Elutriation rate
0.30
Inert bed: 0.2 - 0.3mm Sand 0.85 - 1.0mm Sand 0.2 - 0.3mm Corundum
0.25
X Ca, -
0.20
0.15 0.15
0.12 0.10
0.11
0.05
0.00 0
50
100
150
200
250
Time, min Fig. 2. Degree of calcium conversion as a function of time during FB limestone sulfation tests at 850 °C with different inert beds at 0.73 m s1.
sand or corundum fluidized at 0.73 m s1. XCa is defined as the cumulative (molar) SO2 uptake divided by the initial (molar) calcium inventory. According to this definition, XCa embodies contributions from sulfation of coarse particles as well as of attrited fines. Operating conditions were such that the reactor could not be considered differential with respect to SO2. In fact, SO2 concentration at the exhaust dropped to vanishingly small values at the beginning of the test, to increase thereafter until the inlet concentration was approached in the long term. This feature prevented the assessment of the kinetics of sulfur capture from differential analysis of XCa vs. t curves. The degree of calcium conversion of the sorbent after 240 min was 11–12% with the bed of sand (depending on the sand particle size) and 15% with corundum. These values are significantly smaller than those found for the same limestone under conventional atmospheric FB combustion conditions (about 30% [1]). This is consistent with other results reported in the literature which show that direct sulfation of limestone yields lower conversion degrees in the short term than the sulfation of calcined limestone [11]. It is noteworthy that corundum promotes sulfation of limestone as compared with sand. This is most likely due to more extensive abrasion promoted by corundum which contributes to disclose unconverted calcium for further sulfation.
Fig. 3 reports the sorbent elutriation rate E(t) measured during sulfation at the two fluidization velocities tested. The elutriation rate was normalized with respect to the initial amount of limestone fed to the reactor (W0 = 20 g). A pronounced spike is recorded at the beginning of the test, related to the initial removal of surface asperities (rounding off) from the angular coarse sorbent particles. As rounding off is complete, E(t) drops to a minimum at 5–15 min whence it steadily increases along with the progress of sulfation (see Fig. 2). Notably, the increase of E(t) with the progress of sulfation is opposite to what observed under atmospheric FB combustion conditions, where sulfation hinders attrition and elutriation by strengthening the particle outer layer [1,3]. This difference can be explained by assuming that the sulfur-rich shell formed during direct sulfation of limestone is somewhat softer and more prone to attrition than the original calcium carbonate. Comparison of data reported in Fig. 3A and B indicates that E(t) increases with the gas superficial velocity and decreases with the inert particle size, both features consistent with previous results obtained under different operating conditions [1]. The elutriation rate with corundum is larger than that with sand, consistently with conversion data reported in Fig. 2. It must be underlined, however, that E(t) is only partly related to actual sorbent attrition rate, as part of the attrited material is likely to have been retained in the reactor during the experiments. Elutriation data have been analyzed according to the following equation:
E¼k
0.00030
Inert bed:
A
Inert bed:
B
Sand (0.85-1.0mm) Corundum (0.2-0.3mm)
0.00025 0.00020 0.00015 0.00010
Sand (0.85-1.0mm) Sand (0.2-0.3mm) Corundum (0.2-0.3mm)
0.00025
E/W 0, min -1
E/W 0, min -1
ð1Þ
which relates sorbent elutriation rate to the sorbent loading W, its average particle size d and the gas superficial velocity U. The characteristic velocity U0 in Eq. (1) represents a threshold above which bubble-induced attrition occurs, typically larger than the incipient fluidization velocity Umf of the bed material [12]. Data points in Fig 3A and B have been worked out to yield the instantaneous value of the attrition rate constant k, defined according to Eq. (1) with the approximation W ffi W 0 and d ffi d0 . Values of U0 giving the best fit to the actual data were 0.4 and 0.35 m/s for bed materials in the size ranges 0.3–0.4 mm and 0.85–1.0 mm, respectively. It is remarkable that these values are significantly larger than Umf, a feature that might be related to the high mechanical strength of the sorbent throughout conversion. Values of k obtained with this procedure
0.00030
0.00020 0.00015 0.00010 0.00005
0.00005 0.00000
W U U0 d
0
20
40
60
80
100
Time, min
120
140
160
180
0.00000
0
20
40
60
80
100
120
140
160
180
Time, min
Fig. 3. Limestone elutriation rate as a function of time during FB sulfation tests at 850 °C with different bed materials at gas superficial velocities of 0.51 m s1 (A) and 0.73 m s1 (B).
830
F. Scala, P. Salatino / Fuel 89 (2010) 827–832
1e-8
1e-8
Inert bed: Sand
A
B
U = 0.51m s-1 (0.85-1.0mm)
8e-9
Inert bed: Corundum U = 0.51m s-1
8e-9
U = 0.73m s-1 (0.85-1.0mm) U = 0.73m s-1 (0.2-0.3mm)
6e-9
U = 0.73m s-1
k, -
k, -
6e-9
4e-9
4e-9
2e-9
2e-9
0
0
20
40
60
80
100
120
140
160
0
180
0
20
40
60
Time, min
80
100
120
140
160
180
Time, min
Fig. 4. Sorbent attrition constant k, defined according to Eq. (1), as a function of time during FB sulfation tests at 850 °C at different gas superficial velocities. Bed material: (A) sand and (B) corundum.
are plotted as a function of time in Fig. 4A and B for beds of sand and corundum, respectively. 3.3. Impact fragmentation Samples to be subjected to impact tests were obtained by FB sulfation for 240 min. Sorbent particles were sieved again after sulfation and only the fraction remaining within the 0.4–0.6 mm size interval was subject to impact testing. Figs. 5A and 6A show the probability density function (PDF) of the fragments size for sulfated particles subjected to impact fragmentation tests at different impact velocities for particles pre-processed in beds of sand and corundum, respectively. The same distributions are presented as cumulative fractional mass in Figs. 5B and 6B. By ‘‘fragments” we mean all the particles collected after the impact tests which size falls below the lower limit of the feed size interval: accordingly all the collected particles finer than 0.4 mm were classified as ‘‘fragments” in the context of the present study. Since Figs. 5B and 6B only report the cumulative distributions of the fragments the curves do not end at 1.0, but at a value representing the fractional mass of fragments (the complement to 1.0 being the fractional mass of particles in the size range 0.4–0.6 mm). The PDF was obtained by dividing the fractional mass of particles in a given size bin by the width (in mm) of the size bin.
The extent of fragmentation increases as the impact velocity increases from 10 to 45 m s1. Figs. 5B and 6B show that the fractional amount of fragments is smaller than 10% and 30% of the initial sample mass for sorbent sulfated in sand and corundum, respectively, even at the largest impact velocity of 45 m s1. The extent of impact fragmentation of sorbents pre-processed in a bed of corundum (Fig. 6B) is much larger than that observed for the sand bed (Fig. 5B). Moreover, the effect of increasing v on the extent of particle fragmentation is more pronounced in the former case. Figs. 5A and 6A better highlight the way fragment sizes are distributed. At any impact velocity a dominance of relatively coarse fragments, of size just slightly smaller than 0.4 mm, is observed for the bed of sand (Fig. 5A). On the other hand, a bimodal fragment size distribution is evident as the sorbent is processed in a bed of corundum, with a significant fraction of particles below 100 lm. This difference in the PDF curves is indicative of different fragmentation patterns of the sulfated limestone in the two cases. Fig. 7 summarizes the results obtained with the different samples. Data are reported as the total fractional mass f of fragments (i.e., the total mass of fragments finer than 0.4 mm collected after the impact test) versus impact velocity. The log–log plot was chosen so as to better highlight the establishment of power-law f vs. v relationships (f / vk) that are often used to correlate these data [13]. Inspection of data in Fig. 7 suggests the following features:
2.0
0.30
A
-1
pdf, mm
-1
1.5
-1
cumulative size distribution, -
v=10m s v=17m s-1 v=24m s-1 -1 v=31m s -1 v=38m s -1 v=45m s
1.0
0.5
0.0
B
v=10m s v=17m s-1 -1 v=24m s -1 v=31m s v=38m s-1 v=45m s-1
0.25
0.20
0.15
0.10
0.05
0.00 0.0
0.1
0.2
0.3
particle diameter, mm
0.4
0.5
0.0
0.1
0.2
0.3
0.4
0.5
particle diameter, mm
Fig. 5. Probability density function of particle size (A) and cumulative size distribution (B) of fragments (d < 0.4 mm) collected after impact at different velocities of batches of limestone particles sulfated at 850 °C and 0.73 m s1 with an inert bed of 0.85–1.0 mm sand.
831
F. Scala, P. Salatino / Fuel 89 (2010) 827–832
2.0
0.30
A
-1
cumulative size distribution, -
pdf, mm
-1
1.5
v=10m s v=17m s-1 v=24m s-1 -1 v=31m s -1 v=38m s -1 v=45m s
1.0
0.5
0.0 0.0
v=10m s-1 -1 v=17m s v=24m s-1 -1 v=31m s -1 v=38m s v=45m s-1
0.25
0.20
B
0.15
0.10
0.05
0.00 0.1
0.2
0.3
0.4
0.5
0.0
particle diameter, mm
0.1
0.2
0.3
0.4
0.5
particle diameter, mm
Fig. 6. Probability density function of particle size (A) and cumulative size distribution (B) of fragments (d < 0.4 mm) collected after impact at different velocities of batches of limestone particles sulfated at 850 °C and 0.73 m s1 with an inert bed of 0.2–0.3 mm corundum.
the effect of pre-processing on fragmentation is much emphasized as the impact velocity is raised above 20 m s1; curves relative to the same bed material (corundum vs. sand) tend to overlap at high impact velocity, suggesting that the nature of the bed material plays a much larger role as compared with bed solids particle size and/or gas superficial velocity.
1 -1
0.85 - 1.0mm Sand, 0.51m s
-1
0.2 - 0.3mm Sand, 0.73m s
-1
0.85 - 1.0mm Sand, 0.73m s
0.2 - 0.3mm Corundum, 0.51m s-1 0.2 - 0.3mm Corundum, 0.73m s-1
f, -
0.1
The last point is further supported by data in Fig. 8. Here, the Sauter mean diameters of sorbent particles remaining after impact at different impact velocities decrease to a much larger extent for samples pre-processed in a bed of corundum than in a bed of sand, displaying a minor influence of bed solids size and gas superficial velocity.
0.01
0.001 1
10
100
v, ms
-1
Fig. 7. Fraction of fragmented material as a function of the impact velocity for limestone particles sulfated at 850 °C and subjected to impact fragmentation tests.
Sauter mean diameter, µ m
600
500
400
300
200
0.85 - 1.0mm Sand, 0.51m s-1 0.2 - 0.3mm Sand, 0.73m s-1 0.85 - 1.0mm Sand, 0.73m s-1 0.2 - 0.3mm Corundum, 0.51m s-1 0.2 - 0.3mm Corundum, 0.73m s-1
100
0 0
10
20
30
40
50
-1
v, ms
Fig. 8. Sauter mean diameter of limestone particles sulfated at 850 °C after impact as a function of the impact velocity.
at low impact velocity the extent of fragmentation, as measured by the fragmentation index f, is rather limited, and barely influenced by the pre-processing conditions;
4. Conclusions The propensity of limestone to undergo attrition by surface wear and fragmentation by impact loading under oxyfiring fluidized bed combustion conditions has been assessed by means of a combination of test protocols. Attrition by abrasion is extensive at the very beginning of sorbent sulfation in a fluidized bed, due to initial rounding off of angular raw limestone particles, to decay immediately thereafter. Due to the larger attrition propensity of partly sulfated limestone compared with the raw unconverted one, the progress of sulfation brings about an increase of particle attrition rate. Attrition rate increases with increasing gas superficial velocity and size of bed inert material. The comparison of results of experiments in which limestone was sulfated in a bed of corundum with those in which sulfation occurred in a bed of sand highlight the potential of intentionally promoting attrition to enhance sulfation. Particle sulfation, relatively fast at first, slows down as sulfation proceeds. In the long term, the progress of sulfation appears to be closely linked to the parallel progress of attrition, which makes unconverted calcium accessible for further sulfation. Using corundum as bed material significantly increases sulfur uptake over the same sulfation time. Attrition by impact of pre-processed limestone particles is strongly dependent on particle impact velocity. Limestone samples prepared in a bed of corundum are characterized by a much larger propensity to undergo impact damage when compared with those pre-processed in a bed of sand. This feature is reflected both by the extent and particle size distribution of the debris generated upon each impact.
832
F. Scala, P. Salatino / Fuel 89 (2010) 827–832
Altogether, the phenomenology and mechanisms of limestone attrition and fragmentation under oxyfiring combustion conditions differ significantly from those observed under air-blown atmospheric fluidized bed combustion conditions. Factors that promote attrition may significantly enhance the rate and extent of limestone sulfation. Acknowledgement The experimental support of A. Parlato and G. Somma is gratefully acknowledged. References [1] Scala F, Cammarota A, Chirone R, Salatino P. Comminution of limestone during batch fluidized-bed calcination and sulfation. AIChE J 1997;43:363–73. [2] Di Benedetto A, Salatino P. Modelling attrition of limestone during calcination and sulfation in a fluidized bed reactor. Powder Technol 1998;95:119–28. [3] Scala F, Salatino P, Boerefijn R, Ghadiri M. Attrition of sorbents during fluidized bed calcination and sulphation. Powder Technol 2000;107:153–67. [4] Anthony EJ, Granatstein DL. Sulfation phenomena in fluidized bed combustion systems. Prog Energ Combust Sci 2001;27:215–36.
[5] Shimizu T, Peglow M, Sakuno S, Misawa N, Suzuki N, Ueda H, et al. Effect of attrition on SO2 capture by limestone under pressurized fluidized bed combustion conditions-comparison between a mathematical model of SO2 capture by single limestone particle under attrition condition and SO2 capture in a large-scale PFBC. Chem Eng Sci 2001;56:6719–28. [6] Chen Z, Lim CJ, Grace JR. Study of limestone particle impact attrition. Chem Eng Sci 2007;62:867–77. [7] Scala F, Montagnaro F, Salatino P. Attrition of limestone by impact loading in fluidized beds. Energ Fuel 2007;21:2566–72. [8] Saastamoinen J, Pikkarainen T, Tourunen A, Räsänen M, Jäntti T. Model of fragmentation of limestone particles during thermal shock and calcination in fluidised beds. Powder Technol 2008;187:244–51. [9] Scala F, Salatino P. Flue gas desulfurization under simulated oxyfiring fluidized bed combustion conditions: the influence of limestone attrition and fragmentation. Chem Eng Sci, submitted for publication. [10] Salman AD, Biggs CA, Fua J, Angyal I, Szabó M, Hounslow MJ. An experimental investigation of particle fragmentation using single particle impact studies. Powder Technol 2002;128:36–46. [11] Hu G, Dam-Johansen K, Wedel S, Hansen JP. Review of the direct sulfation reaction of limestone. Prog Energ Combust Sci 2006;32:386–407. [12] Werther J, Reppenhagen J. Attrition. In: Yang W-C, editor. Handbook of fluidization and fluid-particle systems. New York: Marcel Dekker; 2003. p. 239–55. [13] Papadopoulos DG, Ghadiri M. Impact breakage of poly-methylmethacrylate (PMMA) extrudates. 1. Chipping mechanism. Adv Powder Technol 1996;7:183–97.