- Email: [email protected]
Characterization of mineral matter particles in gasification and combustion processes
Fuel 152 (2015) 88–95
Contents lists available at ScienceDirect
Fuel journal homepage: www.elsevier.com/locate/fuel
Characterization of mineral matter particles in gasification and combustion processes Christopher Thiel ⇑, Martin Pohl 1, Sebastian Grahl, Michael Beckmann Institute of Power Engineering, Technische Universität Dresden, George-Bähr-Straße 3b, 01069 Dresden, Germany
h i g h l i g h t s Method to collect particle samples from combustion and gasification processes. Characterization of mineral matter particles depending on process conditions. Information on the deposition probability of mineral matter particles. Information on the precipitation of volatile mineral matter species.
a r t i c l e
i n f o
Article history: Received 15 September 2014 Received in revised form 24 December 2014 Accepted 9 January 2015 Available online 24 January 2015 Keywords: Mineral matter Ash Fossil fuel Biomass Deposition Corrosion
a b s t r a c t Slagging and fouling mechanisms in high temperature solid fuel conversion processes are usually related to the presence of mineral components in the fuel. These mineral components can be extraneous particles or organically bound to the fuel. Both types will be released during the combustion or gasification process. The released mineral particles are diverse and can undergo certain transformation mechanisms along the flue gas path. Common fuel investigation methods rely on bulk ash analysis and other methods that do not take into account the process parameters. It is difficult to predict the deposition and also corrosion behavior of a certain fuel with these laboratory methods alone. An approach with more information about the actual particle situation and where the actual process parameters can be reproduced would be preferable. One way to make the released mineral particles visible, depending on temperature and concentration, is the so called Particle-Wire-Mesh method. Flue gas is extracted from the process through a wire-mesh formed by twilled Dutch weave. The twilled Dutch weave serves as a filter, i.e. particles that are contained in the flue gas will deposit in the pockets or on the wires of the weave. This weave also serves as the sample holder for SEM and EDS analyses. With these analyses the particles can be made visible in the same form as they are extracted from the process. The particles can then be characterized in terms of size, shape, and elemental composition. Size and shape give information about whether the particle has possibly been fused during the combustion process (spherical particles) or not (regularly shaped particles with sharp edges). The location where a particle is found on the wire-mesh can typically be either the wire itself or the pocket. It indicates whether the particle has a probability to stick on furnace walls and heat exchanger surfaces and form a deposit. In order to be found on the wires a particle has to have a sticky surface or impact at a surface that is sticky itself. Particles that are found in the pockets are not sticky. The Particle-Wire-Mesh method of collecting particle samples gives the opportunity to characterize a certain fuel toward the behavior of its mineral content. The metal weave makes it possible to differentiate between single particles. Together with extracting particle samples at different locations in the process, to get the influence of different temperatures, this results in a detailed image of the different species contained in the mineral matter of the fuel. Ó 2015 Elsevier Ltd. All rights reserved.
⇑ Corresponding author. Tel.: +49 351 463 32322. E-mail address: [email protected] (C. Thiel). Present address: ENVERUM Ingenieurgesellschaft für Energie- und Umweltverfahrenstechnik mbH, Hohe Straße 17, 01069 Dresden, Germany. 1
http://dx.doi.org/10.1016/j.fuel.2015.01.025 0016-2361/Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction In spite of the continuing rise of renewable energy sources, such as wind and solar power, the world still relies on solid fuels as one
C. Thiel et al. / Fuel 152 (2015) 88–95
main energy source. Regardless of the fuels origin, fossil or biogenic, all solid fuels contain mineral matter that does not contribute to the energy output but is responsible for fouling and slagging phenomena. The amount of minerals contained in the fuel as well as their composition can vary widely from fuel to fuel. A higher content of minerals in the fuel potentially leads to a higher risk of slagging and fouling in the combustion or gasification facility. But the content of mineral matter in the fuel is not the only criterion contributing to slagging and fouling. Among others, fuel characteristics (e.g. fuel composition, mineral chemistry), process parameters and the design of the reactor2 play a significant role. Mineral matter particles3 in a gasification or combustion process can be very diverse. Each particle can be characterized in regard to origin of the particle (fuel, oxidant, additive or previously deposited and rereleased particle), how the mineral was originally bond in the fuel, its size and shape, the state of matter, chemical composition, and how the particle was released from the fuel or formed out of the gas phase. As the mineral matter particles undergo certain transformation processes (heating and cooling, vapor release, gas phase reactions, fragmentation, heterogeneous or homogeneous condensation, crystallization, etc.) on their way through the reactor, these characteristics may change. The transformation processes can be divided into [1,2] Thermal processes: heating and cooling. Phase change: vapor release and formation of aerosols, fusion and liquefaction, homogeneous or heterogeneous condensation, solidification. Chemical transformations: oxidation, sulfation or other chemical reactions. Mechanical transformations: fragmentation, agglomeration. The characterization of particles at different process locations with different flue gas temperatures creates an image of the process and gives information about these transformations. Conventional fuel investigations rely on laboratory methods. Chemical fractionation, computer controlled scanning electron microscopy (CCSEM) and X-ray fluorescence spectrometry (XRF) are common methods to determine the mineral matter composition in a fuel [3,4]. However, with these methods only the elemental composition can be determined. They give no information about mineral phases. Particularly in regard to corrosion, it is important to know, in which phases and compositions (e.g. salts or silicates) certain elements (e.g. alkaline and alkaline earth metals, sulfur, halogens, lead and zinc) are existent in a gasification or combustion process and how these phases and compositions change along the reaction path. Especially salts are critical for corrosion processes. Ash fusion temperatures can to a certain degree be used to predict the deposition behavior of a specific fuel. However, the method of determining the ash fusion temperatures is not applicable for pulverized fuel fired processes, since the heating rate of the ash samples in this method is only 10 K per minute, significantly lower than in a pulverized fuel fired reactor [5]. All of these fuel analysis methods rely on bulk ash analysis and do not take into account any process specific parameters. Therefore it is difficult to predict the deposition and also corrosion behavior of a certain fuel with these laboratory methods alone. An approach with more information about the actual particle situation and where the actual process parameters can be reproduced would be preferable. The authors, together with CheMin GmbH from Augsburg, Ger2 Reactor refers to either the furnace (in a combustion process) or the gasifier (in a gasification process). 3 Mineral matter particle in the context of the method presented in this paper refers to solid and liquid particles.
89
many, developed a method to collect and characterize particles from gasification and combustion processes [6,7]. This so-called Particle-Wire-Mesh method extracts particles from the flue gas, which then deposit on a metal wire mesh. Characteristics like size and shape can be determined by scanning electron microscopy (SEM) and the elemental composition can be analyzed with energy dispersive X-ray spectroscopy (EDS). The extraction and analysis of particles at different locations and flue gas temperatures creates a detailed image of the particle behavior of that specific fuel. This is a strong basis to understand the combustion or gasification behavior and to predict the deposition behavior in a certain combustion or gasification facility. Problems that are currently being investigated with the Particle-Wire-Mesh method include the impact of changes in fuel or fuel quality on existing furnaces and steam generators, the design of new steam generators for a specific fuel and the impact of combustion optimization, especially NOx reduction mechanisms, on the behavior of mineral matter. 2. The Particle-Wire-Mesh method The Particle-Wire-Mesh method is a method to collect and characterize particle samples at specific locations in the reactor or flue gas duct of a combustion or gasification process. The setup consists of a cylindrical, un-cooled probe and a pressurized air ejector to provide the depression for the extraction flow. The probe has a radial opening which is covered by a metal wire mesh. Flue gas is extracted through the probe, while particles that are contained in the flue gas deposit on the wire mesh. In order to get a single particle layer deposit on the wire mesh, the extraction time is set to 1–5 s, depending on the amount of particles in the flue gas. The short extraction time is necessary to be able to differentiate between single particles in order to characterize them properly. A long extraction time would result in many particles depositing on top of one another, which would make the differentiation between single particles very difficult. The wire mesh used to collect the particle samples is a metal filter (austenitic steel 1.4301) constructed as a twilled Dutch weave with a wire diameter of 20 lm. Four wires form a pocket, as can be seen in Fig. 1. The ideal understanding is that the particles would follow the flue gas flow and accumulate in the pockets. Experiments show however, that the particles not only accumulate in the pockets but also deposit on top of the wires. To be able to adhere to the wire a particle has to impact at a location where the wire has a sticky surface or have a sticky surface itself. A sticky particle surface can be the result of melting or fusion processes, a sticky wire surface can be the result of condensation of mineral vapors out of the gas phase. Also particles can impact at a location on the wire where particles with sticky surfaces have previously impacted. Transport of mineral matter to, and deposition of mineral matter particles at a surface mainly result from impaction and interception, thermophoreses or diffusion mechanisms [3,5,8]. The wire mesh is preheated to flue gas temperature prior to extracting the flue gas. Therefore thermophoreses as a deposition mechanism can be neglected. Fig. 2 shows a basic model of the different ways a particle can deposit on the wire mesh, divided in deposition in the pocket and deposition on the wire. Deposition in the pocket: (a) Particles that can follow the flue gas flow at a change of direction (i.e. fine and very fine particles). (b) Particles that do not need to change direction on their way into the pocket.
90
C. Thiel et al. / Fuel 152 (2015) 88–95
Fig. 1. Empty wire mesh with division into wires and pockets as two different zones for particles to deposit [9].
Fig. 2. Basic scheme of particle deposition mechanisms on a wire mesh [9].
(c) Particles that cannot follow the flue gas flow, bounce off the wire surface and then proceed as in b. (d) Particles that bounce off the wire surface and break into several smaller particles. These smaller particles can then eventually follow the flue gas flow into the pocket. (e) Particles that were formed by condensation or desublimation out of the gas phase can follow the flue gas flow into the pocket. Deposition on the wire: (f) Liquid particles that cannot follow the flue gas flow deposit on the wire. (g) Particles that condense or desublimate directly on the wire surface. (h) An adhesive surface can form around particles by condensation of gaseous species. If these particles impact on the wire, they may stick. (i) Particles that impact on the wire at a location, where other particles have created an adhesive surface by condensation or desublimation. The particle analysis is conducted with a scanning electron microscope (SEM) and electron dispersive X-ray spectrometry (EDS). The morphology and chemical composition of particles are characterized and categorized in terms of size, shape, chemical ele-
ments and location on the wire mesh. Fig. 3 gives examples of particles collected in combustion experiments with Rhenish lignite at an entrained flow reactor at TU Dresden. Size: Particles on the wire mesh can be found in various sizes. We found it practical to categorize the particles as fine (equivalent diameter <1 lm, particles h, i and o in Fig. 3), medium (equivalent diameter between 1 and 10 lm, particles b, d, e, h, i and m) and large particles (equivalent diameter >10 lm, particles a, c, f, g, j, k, l, n, and p). Shape: The particles feature spherical (particles b, c, f, g, h and m), crystalline (not observed with Rhenish lignite, but with other fuels), and irregular shapes (particles a, j, k, l, n and p) throughout all size categories. A spherical shape suggests that this particle has undergone the fused state on its way through the reactor. A crystalline particle can either be a particle that has kept its original shape and has not undergone the fused state or it is the result of a mineral transformation process during combustion or gasification, i.e. crystallization. More information on the particle composition and process temperature is needed to differentiate between these two phenomena. Irregular shapes can occur when fused particles collide or liquid particles solidify directly on the wire surface.
C. Thiel et al. / Fuel 152 (2015) 88–95
91
Fig. 3. Examples of particle samples taken from combustion experiments with Rhenish lignite at an entrained flow reactor.
Chemistry: We measure the elemental particle composition with EDS. The elemental composition of the particles on a wire mesh mainly depends on the fuel composition and the temperature range. Different fuels will yield particles with different compositions, whereas particles collected while testing a certain fuel can be categorized representing certain chemical compounds or phases. Location on the wire mesh: The wire mesh is subdivided into two possible zones for particles to deposit. Particles are either found in the pocket (particles d, g, i, k and o) or on the wire (particles b, f, h and m) (see above, Fig. 2). As previously stated, the location a particle is found on the wire mesh gives information about the stickiness of that particle. Particles on the wire are sticky, whereas particles in the pockets will mainly be non-sticky. However, under certain flow conditions and at small particles sizes, sticky particles can also deposit in the pocket. Also non-sticky particles can deposit on previously deposited sticky particles on the wire.
Characterizing a particle by means of the four features size, shape, chemistry and location on the wire mesh, gives a detailed image of that particle. It also allows suggestions on the particle’s history in terms of release and transformation processes along its way through the reactor.
3. Experimental setup 3.1. Experiments with Rhenish Lignite in an entrained flow reactor We investigated the combustion of a dried Rhenish lignite in an entrained flow reactor with a thermal power output of 50 kW. The setup consists of a down draught swirl burner, an air-cooled combustion chamber, a water cooled heat exchanger and a bag filter. With this configuration we are able to set the process parameters so that the experimental results are transferrable to large scale processes. Fig. 4 shows a schematic of the experimental setup. The test rig is able to operate in oxyfuel-mode, with a full recirculation loop and oxygen supply. However, in this study only exper-
92
C. Thiel et al. / Fuel 152 (2015) 88–95
iments in air fired mode are presented. For experiments in oxyfuelmode see [10,11]. Rhenish lignite is a German low rank coal. Table 1 shows the fuel and bulk ash properties. The mineral matter mainly consists of Ca, Mg, Fe, S and Na. The bulk ash softening temperature was measured to be 1272 °C. All fuel and ash analyses were carried out at our own laboratory. For the experiments the test rig was run at 50 kW thermal load with a coal mass flow of 8 kg/h at 6 volume percent oxygen (dry base) in the flue gas. The coal was transported to the burner with recirculated flue gas. The combustion air was electrically preheated to 350 °C. Particle samples were taken at two different locations of the test rig (see Fig. 4). Sample A was collected at the furnace exit at a flue gas temperature of approximately 700 °C, while sample B was collected downstream the heat exchanger at a flue gas temperature of approximately 200 °C. In order to preheat the wire mesh to
Table 1 Fuel and ash properties of Rhenish lignite and wood pellets. Rhenish lignite
Wood pellets
Mass-% Mass-% Mass-% Mass-% MJ/kg
10.06 49.62 46.54 3.84 26.26
4.82 80.00 19.68 0.32 18.86
Daf Daf Daf Daf Daf
Mass-% Mass-% Mass-% Mass-% Mass-%
66.56 4.84 26.73 1.03 0.83
50.36 6.08 43.03 0.52 0.01
Dry Dry Dry Dry Dry Dry Dry Dry
Mass-% Mass-% Mass-% Mass-% Mass-% Mass-% Mass-% Mass-%
0.32 2.03 10.26 39.43 21.67 17.55 5.38 1.21
5.02 1.03 1.95 31.89 8.66
Proximate analysis Moisture Volatile matter Fixed carbon Ash content LHV
Dry Dry Dry Daf
Ultimate analysis C H O N S Ash composition SiO2 Al2O3 Fe2O3 CaO MgO SO3 Na2O K2O
2.12 12.44
flue gas temperature, the probe is inserted into the flue gas with the opening facing downstream and the extraction valve closed. The valve is then opened for approximately 2 s to force the extraction flow and collect the particles on the wire mesh. 3.2. Experiments with modified wood pellets in a fixed bed semibatch reactor In addition to the experiments with coal in the entrained flow reactor we investigated the combustion behavior of industrial wood pellets in a fixed bed semibatch reactor [9,12]. In a semibatch reactor the drying, volatilization, gasification and combustion processes take place consecutively, partially superimposed. The main objective of these experiments was to investigate the fate of high volatile mineral matter in a biomass fuel. We especially looked at chlorine as a typical example for a high volatile species. In order to get a clear effect we added chlorine containing additives to the wood pellets: NaCl as an example for an inorganically bound chlorine species to a total content of 1.6 mass% and polyvinyl chloride (PVC) as an example for an organically bound chlorine species to a total content of 3.2 mass%. NaCl was added as an aqueous solution and the wood pellets were dried with air again after the addition. For the PVC experiments, floor tiles were cut into small pieces and added to the wood pellets. The setup for these experiments consisted of a fixed bed semibatch reactor with an adjustable grate and a refractory lined furnace (Fig. 5). The furnace was preheated to 800 °C with a natural gas burner to set the ignition conditions on the grate, then fuel was inserted into the retort. By preheating the furnace we achieved a high heating rate of the fuel bed, comparable to a large scale process. The total fuel mass for one batch (wood pellets plus additive) was 11 kg for the experiments with NaCl-additive and 14 kg for the experiments with PVC-additive. Primary air was set to 50 m3/h at standard conditions for both cases. Secondary air was not used. Particle samples were collected at two different locations in the flue gas stream, as shown in Fig. 5, at the transition between the gasification and combustion stage. The particle samples at location C were collected at flue gas temperatures between 540 °C and 560 °C, the particle samples at location D at flue gas temperatures between 300 °C and 320 °C.
Fig. 4. Schematic of the entrained flow reactor with particle sample locations A and B.
C. Thiel et al. / Fuel 152 (2015) 88–95
Fig. 5. Schematic of the fixed bed semibatch reactor with particle sample locations C and D [9].
4. Results and discussion 4.1. Results of experiments with Rhenish lignite in an entrained flow reactor Fig. 6 compares the particles that were collected with the Particle-Wire-Mesh method at the entrained flow reactor at two different locations with different flue gas temperatures. The images were created with a JEOL T 330 A scanning electron microscope. The samples were sputtered with carbon in an EMITECH-K 950Carbon Coater to get the conductive surface for the scanning electron analysis. Both images show spherical and non-spherical particles with equivalent diameters between 1 lm and 20 lm. The particles have deposited in the pockets and on the wires of the wire
93
mesh in both samples. However there are significant differences between both samples. At sample A we found most of the particles on the wires and only some particles in the pockets. The fact that most particles are found on the wires indicates that most particles at this particular location and flue gas temperature have a sticky surface. At sample B on the other hand we found many particles in the pockets. These particles are smaller compared to the particles in sample A and mainly consist of sodium and sulfur. These fine particles are probably sodium sulfate particles that have condensed on their way to location B due to the decreasing flue gas temperature. We barely found any sodium and less sulfur at location A, which indicates that sodium and most of the sulfur was still in the gas phase at this location. The experiments show the different kinds of particles that form during the combustion of Rhenish lignite. Specific particles can be characterized according to the classification scheme described in Section 2. Particle A sticks on the wire (Fig. 6, left) and is made up of several medium-sized merged particles. It mainly consists of calcium and magnesium with some aluminum. Particle B is a large sized, spherical particle and mainly consists of calcium, magnesium and iron with some aluminum and silicon. The location of the particle on the wire is no clear indication of a sticky particle surface since there are some smaller particles underneath particle B. Those particles can provide the adhesive surface for particle B to deposit. Particle C is a medium to large size, spherical particle found on the wire. It mainly consists of silicon with sodium, calcium and potassium and also some aluminum and magnesium. The spherical shape of particles B and C indicate that the furnace temperature was high enough so that these particles could reach their fusion temperature. The agglomeration of particles in the pocket D, mainly small and medium sized particles, contains calcium and sulfur, but no sodium. The location of these particles in the pocket indicates that they are not sticky. The particles on the wire E, also small and medium-sized, mainly spherical particles, contain sodium, calcium and sulfur. Again the spherical shape indicates that these particles had reached the fusion temperature at some point along their way through the furnace. The particles in the pocket and on the wire show a similar elemental composition compared to the furnace bottom ash (Y). At the colder sample position of 200 °C flue gas temperature (Fig. 6, right), particle F is a large, spherical particle, sticking on
Fig. 6. Top: SEM images of particles collected during experiments with Rhenish lignite at the entrained flow reactor; Bottom: EDS results of selected particles (A–J) and XFA results of laboratory ash (X), furnace bottom ash (Y) and bag filter ash (Z).
94
C. Thiel et al. / Fuel 152 (2015) 88–95
the wire that consists mainly of calcium, magnesium and iron. Particle G is similar to particle F except that it has deposited directly in the pocket of the wire mesh. Both particles show the same features and similar elemental composition as particle B that was collected at the higher flue gas temperature, which indicates that these particles have not undergone any transformation processes along their way from the high temperature sample location to the low temperature sample location. Particle H is a medium sized, spherical particle on the wire that mainly consists of sodium and silicon, with small amounts of sulfur, chlorine and potassium. The spherical shape again indicates that this particle has previously reached the fusion temperature. As stated previously, the small particles which fill up the pockets of the wire mesh (I) mainly consist of sodium, calcium and sulfur, with some amounts of potassium and chlorine. These particles should not be a risk for deposition or corrosion and can be removed by particle filters. The particles on the wire (J) are small, mainly spherical and show an elemental composition of calcium, magnesium, sodium and sulfur with traces of potassium and chlorine. These particles have a sticky surface and can potentially contribute to deposition formation in a real combustion or gasification facility. The sticky particle surface that resulted in the particles being deposited on the wire can potentially be due to the condensation of sodium sulfate on these particles [13]. The bulk ash analysis of the fly ash collected in the bag filter (Z) just downstream sample location B shows a similar elemental composition compared to the particles in the pockets and on the wires, with the exception of a smaller sodium content. The experiments showed that particles at a flue gas temperature of 700 °C are more likely to stick at heat exchanger surfaces than at a flue gas temperature of 200 °C. Particles at 700 °C flue gas temperature can contain sulfur, mainly in combination with calcium. At 200 °C flue gas temperature fine particles also contain sodium. The comparison of the results presented in this paper with results obtained in previous experiments at the same test rig show that the Particle-Wire-Mesh method yields reproducible results. Also experiments at the oxyfuel pilot plant in Schwarze Pumpe, which has a thermal capacity of 30 MW, showed that the results
obtained with the particle wire mesh method at pilot scale are transferrable to large scale units [15]. 4.2. Results of experiments with modified wood pellets in a fixed bed semibatch reactor Fig. 7 shows the results of the investigations of the fate of high volatile species in wood pellet combustion. At 540 °C flue gas temperature a crust of medium sized NaCl crystals has formed around the wires of the wire mesh, while at 320 °C the pockets are filled with very fine NaCl particles and the wires are almost free of particles. This behavior suggests that at 540 °C NaCl was dissolved in the flue gas but in an oversaturated state. The wire surface served as an initial nucleus for desublimation, so NaCl crystals formed around the wire. At 320 °C NaCl was in a solid state, so that NaCl particles deposited in the pockets of the wire mesh. The experiments with PVC additive showed similar results (Fig. 7 bottom). At 560 °C flue gas temperature a thin crust of crystals has formed around the wires of the mesh. Also fine particles have filled the pockets. At 300 °C flue gas temperature the pockets are filled with fine particles and the wires are almost free of particles. EDS analysis shows, that the pocket particles from both experiments mainly consist of sodium chloride and potassium chloride. The mass balance for the ash residue in the fuel bed showed that during the PVC experiments more potassium was released than during the experiments with NaCl addition. This indicates that the organically bound chlorine species released during combustion react with the mineral matter of the wood pellets and extract more potassium from the mineral matter than the inorganically bound species. This corresponds to experiments from Sippula et al. [14] with wood combustion and HCl addition that showed that the increased presence of HCl benefits the volatilization of potassium. The experiments with chlorine containing additives showed that chlorine species will precipitate at flue gas temperatures of around 550 °C and form depositions on heat exchanging surfaces. At flue gas temperatures of around 300 °C these particles do not form deposits and therefore should not extensively contribute to fouling processes.
Fig. 7. Top: SEM images of particles collected during experiments with wood pellets and NaCl additive at the fixed bed semibatch reactor; bottom: SEM images of particles collected during experiments with wood pellets and PVC additive at the fixed bed semibatch reactor [15].
C. Thiel et al. / Fuel 152 (2015) 88–95
5. Conclusions With common fuel characterization methods, such as elemental analysis, and the determination of ash fusion temperatures alone it is difficult to characterize the behavior of a fuel in a specific combustion or gasification facility. Especially the prediction of deposition and corrosion behavior needs more information about the mineral matter and its phases and compositions. The particle wire mesh method is a method to collect particle samples from combustion and gasification processes. It creates a snap shot of the particles at a specific location in the flue gas so that the particles can be characterized and their probability of depositing on heat exchanging surfaces can be estimated. To evaluate the deposition probability the Particle-Wire-Mesh method makes use of the wire mesh topography. Particles that deposit on top of wires because of their own adhesive surface are also likely to deposit on other surfaces, while particles that accumulate in the wire mesh pockets do not feature an adhesive surface and are therefore not likely to deposit on heat exchanging surfaces. Another area of application for the particle wire mesh method is to determine where in a combustion or gasification process volatile mineral species precipitate. This is especially important for waste incinerators and other processes where volatile mineral matter such as chlorine species can cause corrosion problems. Experiments were done with lignite at a combustion test rig for pulverized fuels and wood pellets with chlorine containing additives at a fixed bed semibatch reactor. The lignite experiments showed that at 700 °C almost all particles stick to the wire, whereas at 200 °C flue gas temperature many small particles are found in the pockets. The experiments with wood pellets and chlorine containing additives showed that chlorine species will precipitate at flue gas temperatures of around 550 °C and form depositions on heat exchanging surfaces. At flue gas temperatures of around 300 °C these particles do not form deposits and therefore should not extensively contribute to fouling processes. Also the organically bound chlorine species released during combustion react with the mineral matter of the wood pellets and extract more potassium from the mineral matter than the inorganically bound species. All experiments presented in this paper were conducted at a pilot scale. Experiments at the oxyfuel pilot plant in Schwarze Pumpe, which has a thermal capacity of 30 MW, showed, that the results obtained with the Particle-Wire-Mesh method at pilot scale are transferrable to large scale units [15]. The characterization of a fuel with the Particle-Wire-Mesh method in a pilot scale test rig together with established laboratory characterization methods can contribute to understanding the combustion or
95
gasification behavior and estimate the deposition and corrosion probability of a specific fuel. Acknowledgements The authors would like to thank the German Federal Ministry for Economic Affairs and Energy and the German Federal Environmental Foundation for financial support. References [1] Bryers RW. Factors critically affecting fireside deposits in steam generators. In: Gupta RP, Wall TF, Baxter L, editors. Impact of mineral impurities in solid fuel combustion. Boston: Kluwer Academic Publishers; 2002. p. 105–31. [2] Zelkowski J. Kohlecharakterisierung und Kohleverbrennung/Kohle als Brennstoff, Physik und Theorie der Kohleverbrennung, Technik. 2nd ed. Essen (German): VGB PowerTech Service; 2004. [3] Benson SA, Hurley JP, Zygarlicke CJ, Steadman EN, Erickson TA. Predicting ash behavior in utility boilers. Energy Fuels 1993;7. 746–4. [4] Bryers RW. Fireside slagging, fouling, and high-temperature corrosion of heattransfer surface due to impurities in steam-raising fuels. Prog Energy Combust Sci 1996;22:29–120. [5] Gumz W, Kirsch H, Mackowsky MT. Schlackenkunde - Untersuchungen über die Minerale im Brennstoff und ihre Auswirkungen im Kesselbetrieb. 1st ed. Berlin (Göttingen, Heidelberg): Springer; 1958. [6] Spiegel W. Analyse des Rohgases und der Kesselablagerung zur Beurteilung der Ersatzbrennstoffqualität. In: Thomé-Kozmiensky KJ, Beckmann M, editors. Energie aus Abfall – Bd. 1. Neuruppin: TK Verlag Karl Thomé-Kozmiensky; 2006. p. 441–58. [7] Pohl M, Bernhardt D, Ncube S, Beckmann M, Spiegel W, Müller W. Diagnostic methods into the corrosion potential of alternative fuels i.e. biomass and refuse derived fuels. In: Proceedings of the conference on impacts of fuel quality on power production and environment. Saariselkä, Finland; Aug 29– Sep, 2010. [8] Raask E. Mineral impurities in coal combustion: behaviour, problems, and remedial measures. New York: Hemisphere Publishing Corporation; 1985. [9] Pohl M. Methode zur Charakterisierung der freigesetzten mineralischen Bestandteile eines Brennstoffes mit Hilfe der PartikelGitterNetzSonde. Dresden, German: Technische Universität Dresden; 2013 [dissertation]. [10] Thiel C, Pohl M, Beckmann M, Herzog T, Spiegel W. Particle behavior of solid fuels under oxy-fuel conditions. In: Proceedings of the 3rd oxyfuel combustion conference. Ponferrada, Spain; Sep 9–13, 2013. [11] Grahl S, Hiller A, Löser J, Weigl S, Wilhelm R, Beckmann M. Advanced development of the coal-fired oxyfuel process with CO2 separation. In: Proceedings of the 4th international conference on clean coal technologies. Dresden, Germany; May 18–21, 2009. [12] Beckmann M, Scholz R. Modellvorstellungen zum Feststoffumsatz bei Rückständen in Rostfeuerungen. In: Proceedings of GVC–Symposium Abfallwirtschaft – Herausforderung und Chance. Würzburg; Oct 17–19, 1994. p. 251–77. [13] Walsh PM, Sarofim AF, Beer JM. Fouling of convection heat exchangers by lignitic coal ash. Energy Fuels 1992;6:709–15. [14] Sippula O, Lind T, Jokiniemi J. Effects of chlorine and sulphur on particle formation in wood combustion performed in a laboratory scale reactor. Fuel 2008;87:2425–36. [15] Pohl M, Thiel C, Beckmann M, Spiegel W. Brennstoffdiagnose mit Hilfe der PartikelGitterNetzSonde. In: Fritz E, editor. Dampferzeugerkorrosion. Nossen (German): SAXONIA Standortentwicklungs – und – verwaltungsgesellschaft mbH; 2013. p. 157–76.
Contents lists available at ScienceDirect
Fuel journal homepage: www.elsevier.com/locate/fuel
Characterization of mineral matter particles in gasification and combustion processes Christopher Thiel ⇑, Martin Pohl 1, Sebastian Grahl, Michael Beckmann Institute of Power Engineering, Technische Universität Dresden, George-Bähr-Straße 3b, 01069 Dresden, Germany
h i g h l i g h t s Method to collect particle samples from combustion and gasification processes. Characterization of mineral matter particles depending on process conditions. Information on the deposition probability of mineral matter particles. Information on the precipitation of volatile mineral matter species.
a r t i c l e
i n f o
Article history: Received 15 September 2014 Received in revised form 24 December 2014 Accepted 9 January 2015 Available online 24 January 2015 Keywords: Mineral matter Ash Fossil fuel Biomass Deposition Corrosion
a b s t r a c t Slagging and fouling mechanisms in high temperature solid fuel conversion processes are usually related to the presence of mineral components in the fuel. These mineral components can be extraneous particles or organically bound to the fuel. Both types will be released during the combustion or gasification process. The released mineral particles are diverse and can undergo certain transformation mechanisms along the flue gas path. Common fuel investigation methods rely on bulk ash analysis and other methods that do not take into account the process parameters. It is difficult to predict the deposition and also corrosion behavior of a certain fuel with these laboratory methods alone. An approach with more information about the actual particle situation and where the actual process parameters can be reproduced would be preferable. One way to make the released mineral particles visible, depending on temperature and concentration, is the so called Particle-Wire-Mesh method. Flue gas is extracted from the process through a wire-mesh formed by twilled Dutch weave. The twilled Dutch weave serves as a filter, i.e. particles that are contained in the flue gas will deposit in the pockets or on the wires of the weave. This weave also serves as the sample holder for SEM and EDS analyses. With these analyses the particles can be made visible in the same form as they are extracted from the process. The particles can then be characterized in terms of size, shape, and elemental composition. Size and shape give information about whether the particle has possibly been fused during the combustion process (spherical particles) or not (regularly shaped particles with sharp edges). The location where a particle is found on the wire-mesh can typically be either the wire itself or the pocket. It indicates whether the particle has a probability to stick on furnace walls and heat exchanger surfaces and form a deposit. In order to be found on the wires a particle has to have a sticky surface or impact at a surface that is sticky itself. Particles that are found in the pockets are not sticky. The Particle-Wire-Mesh method of collecting particle samples gives the opportunity to characterize a certain fuel toward the behavior of its mineral content. The metal weave makes it possible to differentiate between single particles. Together with extracting particle samples at different locations in the process, to get the influence of different temperatures, this results in a detailed image of the different species contained in the mineral matter of the fuel. Ó 2015 Elsevier Ltd. All rights reserved.
⇑ Corresponding author. Tel.: +49 351 463 32322. E-mail address: [email protected] (C. Thiel). Present address: ENVERUM Ingenieurgesellschaft für Energie- und Umweltverfahrenstechnik mbH, Hohe Straße 17, 01069 Dresden, Germany. 1
http://dx.doi.org/10.1016/j.fuel.2015.01.025 0016-2361/Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction In spite of the continuing rise of renewable energy sources, such as wind and solar power, the world still relies on solid fuels as one
C. Thiel et al. / Fuel 152 (2015) 88–95
main energy source. Regardless of the fuels origin, fossil or biogenic, all solid fuels contain mineral matter that does not contribute to the energy output but is responsible for fouling and slagging phenomena. The amount of minerals contained in the fuel as well as their composition can vary widely from fuel to fuel. A higher content of minerals in the fuel potentially leads to a higher risk of slagging and fouling in the combustion or gasification facility. But the content of mineral matter in the fuel is not the only criterion contributing to slagging and fouling. Among others, fuel characteristics (e.g. fuel composition, mineral chemistry), process parameters and the design of the reactor2 play a significant role. Mineral matter particles3 in a gasification or combustion process can be very diverse. Each particle can be characterized in regard to origin of the particle (fuel, oxidant, additive or previously deposited and rereleased particle), how the mineral was originally bond in the fuel, its size and shape, the state of matter, chemical composition, and how the particle was released from the fuel or formed out of the gas phase. As the mineral matter particles undergo certain transformation processes (heating and cooling, vapor release, gas phase reactions, fragmentation, heterogeneous or homogeneous condensation, crystallization, etc.) on their way through the reactor, these characteristics may change. The transformation processes can be divided into [1,2] Thermal processes: heating and cooling. Phase change: vapor release and formation of aerosols, fusion and liquefaction, homogeneous or heterogeneous condensation, solidification. Chemical transformations: oxidation, sulfation or other chemical reactions. Mechanical transformations: fragmentation, agglomeration. The characterization of particles at different process locations with different flue gas temperatures creates an image of the process and gives information about these transformations. Conventional fuel investigations rely on laboratory methods. Chemical fractionation, computer controlled scanning electron microscopy (CCSEM) and X-ray fluorescence spectrometry (XRF) are common methods to determine the mineral matter composition in a fuel [3,4]. However, with these methods only the elemental composition can be determined. They give no information about mineral phases. Particularly in regard to corrosion, it is important to know, in which phases and compositions (e.g. salts or silicates) certain elements (e.g. alkaline and alkaline earth metals, sulfur, halogens, lead and zinc) are existent in a gasification or combustion process and how these phases and compositions change along the reaction path. Especially salts are critical for corrosion processes. Ash fusion temperatures can to a certain degree be used to predict the deposition behavior of a specific fuel. However, the method of determining the ash fusion temperatures is not applicable for pulverized fuel fired processes, since the heating rate of the ash samples in this method is only 10 K per minute, significantly lower than in a pulverized fuel fired reactor [5]. All of these fuel analysis methods rely on bulk ash analysis and do not take into account any process specific parameters. Therefore it is difficult to predict the deposition and also corrosion behavior of a certain fuel with these laboratory methods alone. An approach with more information about the actual particle situation and where the actual process parameters can be reproduced would be preferable. The authors, together with CheMin GmbH from Augsburg, Ger2 Reactor refers to either the furnace (in a combustion process) or the gasifier (in a gasification process). 3 Mineral matter particle in the context of the method presented in this paper refers to solid and liquid particles.
89
many, developed a method to collect and characterize particles from gasification and combustion processes [6,7]. This so-called Particle-Wire-Mesh method extracts particles from the flue gas, which then deposit on a metal wire mesh. Characteristics like size and shape can be determined by scanning electron microscopy (SEM) and the elemental composition can be analyzed with energy dispersive X-ray spectroscopy (EDS). The extraction and analysis of particles at different locations and flue gas temperatures creates a detailed image of the particle behavior of that specific fuel. This is a strong basis to understand the combustion or gasification behavior and to predict the deposition behavior in a certain combustion or gasification facility. Problems that are currently being investigated with the Particle-Wire-Mesh method include the impact of changes in fuel or fuel quality on existing furnaces and steam generators, the design of new steam generators for a specific fuel and the impact of combustion optimization, especially NOx reduction mechanisms, on the behavior of mineral matter. 2. The Particle-Wire-Mesh method The Particle-Wire-Mesh method is a method to collect and characterize particle samples at specific locations in the reactor or flue gas duct of a combustion or gasification process. The setup consists of a cylindrical, un-cooled probe and a pressurized air ejector to provide the depression for the extraction flow. The probe has a radial opening which is covered by a metal wire mesh. Flue gas is extracted through the probe, while particles that are contained in the flue gas deposit on the wire mesh. In order to get a single particle layer deposit on the wire mesh, the extraction time is set to 1–5 s, depending on the amount of particles in the flue gas. The short extraction time is necessary to be able to differentiate between single particles in order to characterize them properly. A long extraction time would result in many particles depositing on top of one another, which would make the differentiation between single particles very difficult. The wire mesh used to collect the particle samples is a metal filter (austenitic steel 1.4301) constructed as a twilled Dutch weave with a wire diameter of 20 lm. Four wires form a pocket, as can be seen in Fig. 1. The ideal understanding is that the particles would follow the flue gas flow and accumulate in the pockets. Experiments show however, that the particles not only accumulate in the pockets but also deposit on top of the wires. To be able to adhere to the wire a particle has to impact at a location where the wire has a sticky surface or have a sticky surface itself. A sticky particle surface can be the result of melting or fusion processes, a sticky wire surface can be the result of condensation of mineral vapors out of the gas phase. Also particles can impact at a location on the wire where particles with sticky surfaces have previously impacted. Transport of mineral matter to, and deposition of mineral matter particles at a surface mainly result from impaction and interception, thermophoreses or diffusion mechanisms [3,5,8]. The wire mesh is preheated to flue gas temperature prior to extracting the flue gas. Therefore thermophoreses as a deposition mechanism can be neglected. Fig. 2 shows a basic model of the different ways a particle can deposit on the wire mesh, divided in deposition in the pocket and deposition on the wire. Deposition in the pocket: (a) Particles that can follow the flue gas flow at a change of direction (i.e. fine and very fine particles). (b) Particles that do not need to change direction on their way into the pocket.
90
C. Thiel et al. / Fuel 152 (2015) 88–95
Fig. 1. Empty wire mesh with division into wires and pockets as two different zones for particles to deposit [9].
Fig. 2. Basic scheme of particle deposition mechanisms on a wire mesh [9].
(c) Particles that cannot follow the flue gas flow, bounce off the wire surface and then proceed as in b. (d) Particles that bounce off the wire surface and break into several smaller particles. These smaller particles can then eventually follow the flue gas flow into the pocket. (e) Particles that were formed by condensation or desublimation out of the gas phase can follow the flue gas flow into the pocket. Deposition on the wire: (f) Liquid particles that cannot follow the flue gas flow deposit on the wire. (g) Particles that condense or desublimate directly on the wire surface. (h) An adhesive surface can form around particles by condensation of gaseous species. If these particles impact on the wire, they may stick. (i) Particles that impact on the wire at a location, where other particles have created an adhesive surface by condensation or desublimation. The particle analysis is conducted with a scanning electron microscope (SEM) and electron dispersive X-ray spectrometry (EDS). The morphology and chemical composition of particles are characterized and categorized in terms of size, shape, chemical ele-
ments and location on the wire mesh. Fig. 3 gives examples of particles collected in combustion experiments with Rhenish lignite at an entrained flow reactor at TU Dresden. Size: Particles on the wire mesh can be found in various sizes. We found it practical to categorize the particles as fine (equivalent diameter <1 lm, particles h, i and o in Fig. 3), medium (equivalent diameter between 1 and 10 lm, particles b, d, e, h, i and m) and large particles (equivalent diameter >10 lm, particles a, c, f, g, j, k, l, n, and p). Shape: The particles feature spherical (particles b, c, f, g, h and m), crystalline (not observed with Rhenish lignite, but with other fuels), and irregular shapes (particles a, j, k, l, n and p) throughout all size categories. A spherical shape suggests that this particle has undergone the fused state on its way through the reactor. A crystalline particle can either be a particle that has kept its original shape and has not undergone the fused state or it is the result of a mineral transformation process during combustion or gasification, i.e. crystallization. More information on the particle composition and process temperature is needed to differentiate between these two phenomena. Irregular shapes can occur when fused particles collide or liquid particles solidify directly on the wire surface.
C. Thiel et al. / Fuel 152 (2015) 88–95
91
Fig. 3. Examples of particle samples taken from combustion experiments with Rhenish lignite at an entrained flow reactor.
Chemistry: We measure the elemental particle composition with EDS. The elemental composition of the particles on a wire mesh mainly depends on the fuel composition and the temperature range. Different fuels will yield particles with different compositions, whereas particles collected while testing a certain fuel can be categorized representing certain chemical compounds or phases. Location on the wire mesh: The wire mesh is subdivided into two possible zones for particles to deposit. Particles are either found in the pocket (particles d, g, i, k and o) or on the wire (particles b, f, h and m) (see above, Fig. 2). As previously stated, the location a particle is found on the wire mesh gives information about the stickiness of that particle. Particles on the wire are sticky, whereas particles in the pockets will mainly be non-sticky. However, under certain flow conditions and at small particles sizes, sticky particles can also deposit in the pocket. Also non-sticky particles can deposit on previously deposited sticky particles on the wire.
Characterizing a particle by means of the four features size, shape, chemistry and location on the wire mesh, gives a detailed image of that particle. It also allows suggestions on the particle’s history in terms of release and transformation processes along its way through the reactor.
3. Experimental setup 3.1. Experiments with Rhenish Lignite in an entrained flow reactor We investigated the combustion of a dried Rhenish lignite in an entrained flow reactor with a thermal power output of 50 kW. The setup consists of a down draught swirl burner, an air-cooled combustion chamber, a water cooled heat exchanger and a bag filter. With this configuration we are able to set the process parameters so that the experimental results are transferrable to large scale processes. Fig. 4 shows a schematic of the experimental setup. The test rig is able to operate in oxyfuel-mode, with a full recirculation loop and oxygen supply. However, in this study only exper-
92
C. Thiel et al. / Fuel 152 (2015) 88–95
iments in air fired mode are presented. For experiments in oxyfuelmode see [10,11]. Rhenish lignite is a German low rank coal. Table 1 shows the fuel and bulk ash properties. The mineral matter mainly consists of Ca, Mg, Fe, S and Na. The bulk ash softening temperature was measured to be 1272 °C. All fuel and ash analyses were carried out at our own laboratory. For the experiments the test rig was run at 50 kW thermal load with a coal mass flow of 8 kg/h at 6 volume percent oxygen (dry base) in the flue gas. The coal was transported to the burner with recirculated flue gas. The combustion air was electrically preheated to 350 °C. Particle samples were taken at two different locations of the test rig (see Fig. 4). Sample A was collected at the furnace exit at a flue gas temperature of approximately 700 °C, while sample B was collected downstream the heat exchanger at a flue gas temperature of approximately 200 °C. In order to preheat the wire mesh to
Table 1 Fuel and ash properties of Rhenish lignite and wood pellets. Rhenish lignite
Wood pellets
Mass-% Mass-% Mass-% Mass-% MJ/kg
10.06 49.62 46.54 3.84 26.26
4.82 80.00 19.68 0.32 18.86
Daf Daf Daf Daf Daf
Mass-% Mass-% Mass-% Mass-% Mass-%
66.56 4.84 26.73 1.03 0.83
50.36 6.08 43.03 0.52 0.01
Dry Dry Dry Dry Dry Dry Dry Dry
Mass-% Mass-% Mass-% Mass-% Mass-% Mass-% Mass-% Mass-%
0.32 2.03 10.26 39.43 21.67 17.55 5.38 1.21
5.02 1.03 1.95 31.89 8.66
Proximate analysis Moisture Volatile matter Fixed carbon Ash content LHV
Dry Dry Dry Daf
Ultimate analysis C H O N S Ash composition SiO2 Al2O3 Fe2O3 CaO MgO SO3 Na2O K2O
2.12 12.44
flue gas temperature, the probe is inserted into the flue gas with the opening facing downstream and the extraction valve closed. The valve is then opened for approximately 2 s to force the extraction flow and collect the particles on the wire mesh. 3.2. Experiments with modified wood pellets in a fixed bed semibatch reactor In addition to the experiments with coal in the entrained flow reactor we investigated the combustion behavior of industrial wood pellets in a fixed bed semibatch reactor [9,12]. In a semibatch reactor the drying, volatilization, gasification and combustion processes take place consecutively, partially superimposed. The main objective of these experiments was to investigate the fate of high volatile mineral matter in a biomass fuel. We especially looked at chlorine as a typical example for a high volatile species. In order to get a clear effect we added chlorine containing additives to the wood pellets: NaCl as an example for an inorganically bound chlorine species to a total content of 1.6 mass% and polyvinyl chloride (PVC) as an example for an organically bound chlorine species to a total content of 3.2 mass%. NaCl was added as an aqueous solution and the wood pellets were dried with air again after the addition. For the PVC experiments, floor tiles were cut into small pieces and added to the wood pellets. The setup for these experiments consisted of a fixed bed semibatch reactor with an adjustable grate and a refractory lined furnace (Fig. 5). The furnace was preheated to 800 °C with a natural gas burner to set the ignition conditions on the grate, then fuel was inserted into the retort. By preheating the furnace we achieved a high heating rate of the fuel bed, comparable to a large scale process. The total fuel mass for one batch (wood pellets plus additive) was 11 kg for the experiments with NaCl-additive and 14 kg for the experiments with PVC-additive. Primary air was set to 50 m3/h at standard conditions for both cases. Secondary air was not used. Particle samples were collected at two different locations in the flue gas stream, as shown in Fig. 5, at the transition between the gasification and combustion stage. The particle samples at location C were collected at flue gas temperatures between 540 °C and 560 °C, the particle samples at location D at flue gas temperatures between 300 °C and 320 °C.
Fig. 4. Schematic of the entrained flow reactor with particle sample locations A and B.
C. Thiel et al. / Fuel 152 (2015) 88–95
Fig. 5. Schematic of the fixed bed semibatch reactor with particle sample locations C and D [9].
4. Results and discussion 4.1. Results of experiments with Rhenish lignite in an entrained flow reactor Fig. 6 compares the particles that were collected with the Particle-Wire-Mesh method at the entrained flow reactor at two different locations with different flue gas temperatures. The images were created with a JEOL T 330 A scanning electron microscope. The samples were sputtered with carbon in an EMITECH-K 950Carbon Coater to get the conductive surface for the scanning electron analysis. Both images show spherical and non-spherical particles with equivalent diameters between 1 lm and 20 lm. The particles have deposited in the pockets and on the wires of the wire
93
mesh in both samples. However there are significant differences between both samples. At sample A we found most of the particles on the wires and only some particles in the pockets. The fact that most particles are found on the wires indicates that most particles at this particular location and flue gas temperature have a sticky surface. At sample B on the other hand we found many particles in the pockets. These particles are smaller compared to the particles in sample A and mainly consist of sodium and sulfur. These fine particles are probably sodium sulfate particles that have condensed on their way to location B due to the decreasing flue gas temperature. We barely found any sodium and less sulfur at location A, which indicates that sodium and most of the sulfur was still in the gas phase at this location. The experiments show the different kinds of particles that form during the combustion of Rhenish lignite. Specific particles can be characterized according to the classification scheme described in Section 2. Particle A sticks on the wire (Fig. 6, left) and is made up of several medium-sized merged particles. It mainly consists of calcium and magnesium with some aluminum. Particle B is a large sized, spherical particle and mainly consists of calcium, magnesium and iron with some aluminum and silicon. The location of the particle on the wire is no clear indication of a sticky particle surface since there are some smaller particles underneath particle B. Those particles can provide the adhesive surface for particle B to deposit. Particle C is a medium to large size, spherical particle found on the wire. It mainly consists of silicon with sodium, calcium and potassium and also some aluminum and magnesium. The spherical shape of particles B and C indicate that the furnace temperature was high enough so that these particles could reach their fusion temperature. The agglomeration of particles in the pocket D, mainly small and medium sized particles, contains calcium and sulfur, but no sodium. The location of these particles in the pocket indicates that they are not sticky. The particles on the wire E, also small and medium-sized, mainly spherical particles, contain sodium, calcium and sulfur. Again the spherical shape indicates that these particles had reached the fusion temperature at some point along their way through the furnace. The particles in the pocket and on the wire show a similar elemental composition compared to the furnace bottom ash (Y). At the colder sample position of 200 °C flue gas temperature (Fig. 6, right), particle F is a large, spherical particle, sticking on
Fig. 6. Top: SEM images of particles collected during experiments with Rhenish lignite at the entrained flow reactor; Bottom: EDS results of selected particles (A–J) and XFA results of laboratory ash (X), furnace bottom ash (Y) and bag filter ash (Z).
94
C. Thiel et al. / Fuel 152 (2015) 88–95
the wire that consists mainly of calcium, magnesium and iron. Particle G is similar to particle F except that it has deposited directly in the pocket of the wire mesh. Both particles show the same features and similar elemental composition as particle B that was collected at the higher flue gas temperature, which indicates that these particles have not undergone any transformation processes along their way from the high temperature sample location to the low temperature sample location. Particle H is a medium sized, spherical particle on the wire that mainly consists of sodium and silicon, with small amounts of sulfur, chlorine and potassium. The spherical shape again indicates that this particle has previously reached the fusion temperature. As stated previously, the small particles which fill up the pockets of the wire mesh (I) mainly consist of sodium, calcium and sulfur, with some amounts of potassium and chlorine. These particles should not be a risk for deposition or corrosion and can be removed by particle filters. The particles on the wire (J) are small, mainly spherical and show an elemental composition of calcium, magnesium, sodium and sulfur with traces of potassium and chlorine. These particles have a sticky surface and can potentially contribute to deposition formation in a real combustion or gasification facility. The sticky particle surface that resulted in the particles being deposited on the wire can potentially be due to the condensation of sodium sulfate on these particles [13]. The bulk ash analysis of the fly ash collected in the bag filter (Z) just downstream sample location B shows a similar elemental composition compared to the particles in the pockets and on the wires, with the exception of a smaller sodium content. The experiments showed that particles at a flue gas temperature of 700 °C are more likely to stick at heat exchanger surfaces than at a flue gas temperature of 200 °C. Particles at 700 °C flue gas temperature can contain sulfur, mainly in combination with calcium. At 200 °C flue gas temperature fine particles also contain sodium. The comparison of the results presented in this paper with results obtained in previous experiments at the same test rig show that the Particle-Wire-Mesh method yields reproducible results. Also experiments at the oxyfuel pilot plant in Schwarze Pumpe, which has a thermal capacity of 30 MW, showed that the results
obtained with the particle wire mesh method at pilot scale are transferrable to large scale units [15]. 4.2. Results of experiments with modified wood pellets in a fixed bed semibatch reactor Fig. 7 shows the results of the investigations of the fate of high volatile species in wood pellet combustion. At 540 °C flue gas temperature a crust of medium sized NaCl crystals has formed around the wires of the wire mesh, while at 320 °C the pockets are filled with very fine NaCl particles and the wires are almost free of particles. This behavior suggests that at 540 °C NaCl was dissolved in the flue gas but in an oversaturated state. The wire surface served as an initial nucleus for desublimation, so NaCl crystals formed around the wire. At 320 °C NaCl was in a solid state, so that NaCl particles deposited in the pockets of the wire mesh. The experiments with PVC additive showed similar results (Fig. 7 bottom). At 560 °C flue gas temperature a thin crust of crystals has formed around the wires of the mesh. Also fine particles have filled the pockets. At 300 °C flue gas temperature the pockets are filled with fine particles and the wires are almost free of particles. EDS analysis shows, that the pocket particles from both experiments mainly consist of sodium chloride and potassium chloride. The mass balance for the ash residue in the fuel bed showed that during the PVC experiments more potassium was released than during the experiments with NaCl addition. This indicates that the organically bound chlorine species released during combustion react with the mineral matter of the wood pellets and extract more potassium from the mineral matter than the inorganically bound species. This corresponds to experiments from Sippula et al. [14] with wood combustion and HCl addition that showed that the increased presence of HCl benefits the volatilization of potassium. The experiments with chlorine containing additives showed that chlorine species will precipitate at flue gas temperatures of around 550 °C and form depositions on heat exchanging surfaces. At flue gas temperatures of around 300 °C these particles do not form deposits and therefore should not extensively contribute to fouling processes.
Fig. 7. Top: SEM images of particles collected during experiments with wood pellets and NaCl additive at the fixed bed semibatch reactor; bottom: SEM images of particles collected during experiments with wood pellets and PVC additive at the fixed bed semibatch reactor [15].
C. Thiel et al. / Fuel 152 (2015) 88–95
5. Conclusions With common fuel characterization methods, such as elemental analysis, and the determination of ash fusion temperatures alone it is difficult to characterize the behavior of a fuel in a specific combustion or gasification facility. Especially the prediction of deposition and corrosion behavior needs more information about the mineral matter and its phases and compositions. The particle wire mesh method is a method to collect particle samples from combustion and gasification processes. It creates a snap shot of the particles at a specific location in the flue gas so that the particles can be characterized and their probability of depositing on heat exchanging surfaces can be estimated. To evaluate the deposition probability the Particle-Wire-Mesh method makes use of the wire mesh topography. Particles that deposit on top of wires because of their own adhesive surface are also likely to deposit on other surfaces, while particles that accumulate in the wire mesh pockets do not feature an adhesive surface and are therefore not likely to deposit on heat exchanging surfaces. Another area of application for the particle wire mesh method is to determine where in a combustion or gasification process volatile mineral species precipitate. This is especially important for waste incinerators and other processes where volatile mineral matter such as chlorine species can cause corrosion problems. Experiments were done with lignite at a combustion test rig for pulverized fuels and wood pellets with chlorine containing additives at a fixed bed semibatch reactor. The lignite experiments showed that at 700 °C almost all particles stick to the wire, whereas at 200 °C flue gas temperature many small particles are found in the pockets. The experiments with wood pellets and chlorine containing additives showed that chlorine species will precipitate at flue gas temperatures of around 550 °C and form depositions on heat exchanging surfaces. At flue gas temperatures of around 300 °C these particles do not form deposits and therefore should not extensively contribute to fouling processes. Also the organically bound chlorine species released during combustion react with the mineral matter of the wood pellets and extract more potassium from the mineral matter than the inorganically bound species. All experiments presented in this paper were conducted at a pilot scale. Experiments at the oxyfuel pilot plant in Schwarze Pumpe, which has a thermal capacity of 30 MW, showed, that the results obtained with the Particle-Wire-Mesh method at pilot scale are transferrable to large scale units [15]. The characterization of a fuel with the Particle-Wire-Mesh method in a pilot scale test rig together with established laboratory characterization methods can contribute to understanding the combustion or
95
gasification behavior and estimate the deposition and corrosion probability of a specific fuel. Acknowledgements The authors would like to thank the German Federal Ministry for Economic Affairs and Energy and the German Federal Environmental Foundation for financial support. References [1] Bryers RW. Factors critically affecting fireside deposits in steam generators. In: Gupta RP, Wall TF, Baxter L, editors. Impact of mineral impurities in solid fuel combustion. Boston: Kluwer Academic Publishers; 2002. p. 105–31. [2] Zelkowski J. Kohlecharakterisierung und Kohleverbrennung/Kohle als Brennstoff, Physik und Theorie der Kohleverbrennung, Technik. 2nd ed. Essen (German): VGB PowerTech Service; 2004. [3] Benson SA, Hurley JP, Zygarlicke CJ, Steadman EN, Erickson TA. Predicting ash behavior in utility boilers. Energy Fuels 1993;7. 746–4. [4] Bryers RW. Fireside slagging, fouling, and high-temperature corrosion of heattransfer surface due to impurities in steam-raising fuels. Prog Energy Combust Sci 1996;22:29–120. [5] Gumz W, Kirsch H, Mackowsky MT. Schlackenkunde - Untersuchungen über die Minerale im Brennstoff und ihre Auswirkungen im Kesselbetrieb. 1st ed. Berlin (Göttingen, Heidelberg): Springer; 1958. [6] Spiegel W. Analyse des Rohgases und der Kesselablagerung zur Beurteilung der Ersatzbrennstoffqualität. In: Thomé-Kozmiensky KJ, Beckmann M, editors. Energie aus Abfall – Bd. 1. Neuruppin: TK Verlag Karl Thomé-Kozmiensky; 2006. p. 441–58. [7] Pohl M, Bernhardt D, Ncube S, Beckmann M, Spiegel W, Müller W. Diagnostic methods into the corrosion potential of alternative fuels i.e. biomass and refuse derived fuels. In: Proceedings of the conference on impacts of fuel quality on power production and environment. Saariselkä, Finland; Aug 29– Sep, 2010. [8] Raask E. Mineral impurities in coal combustion: behaviour, problems, and remedial measures. New York: Hemisphere Publishing Corporation; 1985. [9] Pohl M. Methode zur Charakterisierung der freigesetzten mineralischen Bestandteile eines Brennstoffes mit Hilfe der PartikelGitterNetzSonde. Dresden, German: Technische Universität Dresden; 2013 [dissertation]. [10] Thiel C, Pohl M, Beckmann M, Herzog T, Spiegel W. Particle behavior of solid fuels under oxy-fuel conditions. In: Proceedings of the 3rd oxyfuel combustion conference. Ponferrada, Spain; Sep 9–13, 2013. [11] Grahl S, Hiller A, Löser J, Weigl S, Wilhelm R, Beckmann M. Advanced development of the coal-fired oxyfuel process with CO2 separation. In: Proceedings of the 4th international conference on clean coal technologies. Dresden, Germany; May 18–21, 2009. [12] Beckmann M, Scholz R. Modellvorstellungen zum Feststoffumsatz bei Rückständen in Rostfeuerungen. In: Proceedings of GVC–Symposium Abfallwirtschaft – Herausforderung und Chance. Würzburg; Oct 17–19, 1994. p. 251–77. [13] Walsh PM, Sarofim AF, Beer JM. Fouling of convection heat exchangers by lignitic coal ash. Energy Fuels 1992;6:709–15. [14] Sippula O, Lind T, Jokiniemi J. Effects of chlorine and sulphur on particle formation in wood combustion performed in a laboratory scale reactor. Fuel 2008;87:2425–36. [15] Pohl M, Thiel C, Beckmann M, Spiegel W. Brennstoffdiagnose mit Hilfe der PartikelGitterNetzSonde. In: Fritz E, editor. Dampferzeugerkorrosion. Nossen (German): SAXONIA Standortentwicklungs – und – verwaltungsgesellschaft mbH; 2013. p. 157–76.