2948
Energy & Fuels 2008, 22, 2948–2954
Viscosity of Ashes from Energy Production and Municipal Solid Waste Handling: A Comparative Study between Two Different Experimental Setups S. Arvelakis,*,† F. J. Frandsen,† B. Folkedahl,‡ and J. Hurley‡ CHEC Research Group, Department of Chemical Engineering, Technical UniVersity of Denmark (DTU), Lyngby 2800, Denmark, and Energy and EnVironmental Research Center, UniVersity of North Dakota, 15 North 23rd Street, Stop 9018, Grand Forks, North Dakota 58202-9018 ReceiVed February 10, 2008. ReVised Manuscript ReceiVed June 15, 2008
This paper discusses the viscosity characteristics of ash fractions produced from the co-combustion of coal and biomass in a pilot-scale pulverized fuel (PF) boiler and from the incineration of municipal solid waste (MSW) in a Danish incinerator that were determined using the high-temperature rotational viscometer method. Two different setups in the United States and Denmark were purchased by the same company, HAAKE, but different versions of the basic instrument were used in the study to determine the reproducibility of the method. The two sets of measurements show that the method generates, in principle, reproducible results. The comparison of the generated results as well as the operation and measuring capabilities of the two setups provide useful information regarding the interpretation of the viscosity data as well as the capabilities and the limitations of the method and the specific setups.
Introduction To reduce corrosion and deposition problems, the melting behavior as well as the flowing characteristics of the produced ash deposits must be known as a function of temperature. The flow characteristics of the generated ash slags/deposits depend upon their viscosity. The viscosity of ash melts is dependent upon chemical composition, oxygen level, and temperature. Viscosity is a non-equilibrium property, a measure of the resistance of a fluid toward motion.1-3 It has been shown that the strength of ash-fouling deposits is inversely proportional to the viscosity of the liquid phases present on them. Several publications have pointed out the importance of measuring and predicting viscosities of coal ash slags and related melt phases to ensure a trouble-free operation of combustors and gasifiers. The modeling and prediction of viscosity are also important in the development of comprehensive models of ash behavior in boilers.4-13 In this study, the viscosity characteristics of several
ash fractions from a municipal solid waste (MSW) incinerator in Denmark and the co-combustion of coal and straw in a pilotscale pulverized fuel (PF) boiler were studied using two different high-temperature viscometers. The produced results provide a basis for a better understanding of the flow characteristics of the fly ash generated during the biomass co-combustion with coal as well as the MSW incineration. The comparison of the generated results and the operation and measuring capabilities of the two viscometers provide useful information for the interpretation of the viscosity data and the capabilities and limitations of the measuring method and the viscometers. Additionally, the experimental data may be used as entry data for the development and validation of new viscosity models to describe the flow behavior of ash materials as well to model the ash deposit buildup and thus provide valuable data toward this direction. Materials and Methods
* To whom correspondence should be addressed. Telephone: +4545252883. Fax: +45-45882258. E-mail:
[email protected]. † Technical University of Denmark (DTU). ‡ University of North Dakota. (1) Kestin, J.; Wakeham, W. A. The measurement of viscosity. In Transport Properties of Fluids, Thermal ConductiVity, Viscosity, and Diffusion Coefficient; Ho, C. Y., Ed.; Hemisphere: New York, 1988; Chapter 4, pp 73-147. (2) Viscosity. In Perry’s Chemical Engineers Handbook, 6th ed.; Perry, R. H., Green, D. W., Maloney, J. O., Eds.; McGraw-Hill: Singapore, 1984; Chapters 3 and 5, p 3-278-82. (3) Reid, R. C.; Prausnitz, J. M.; Poling, B. E. In The Properties of Gases and Liquids; Reid, R. C., Prausnitz, J. M., Poling, B. E., Eds.; McGraw-Hill: Singapore, 1988; p 388. (4) Vargas, S.; Frandsen, F. J.; Dam-Johansen, K. Prog. Energy Combust. 2001, 27, 237. (5) Vuthaluru, H. B.; Domazetis, G.; Wall, T. F.; Vleeskens, J. M. Fuel Process. Technol. 1996, 46, 117. (6) Folkedahl, B. C.; Schobert, H. H. Energy Fuels 2005, 19, 208. (7) Hurst, H. J.; Nowak, F.; Patterson, J. H. Fuel 1999, 78, 439. (8) Nowok, J. W.; Hurley, J. P.; Steadman, E. N. The impact of ash deposition on coal fired plants 1993. June 21-25, 1994.
The four co-combustion ash samples (OP18 cyclone, OP19 cyclone, OP19 filter, and OP20 filter) used in this study were produced from the co-combustion of coal [German brown coal (RBF), bituminous coal Go¨ttelborn] and wheat straw at cocombustion shares of 25% for the OP18 and OP20 and 50% for the OP19 runs at the 0.5 MWth pulverized fuel combustion (PFC) rig at IVD Research Institute as a part of the EU Joule III research project Operational Problems, Trace Emissions and By-Product (9) Nowok, J. W.; Benson, S. A. Inorganic transformations and ash deposition during combustion 1991. March 10-15, 1991. (10) Arvelakis, S.; Frandsen, F. J. Viscosity characteristics of ashes from the co-combustion of coal and straw in a pilot scale PF boiler. Presented at the 7th European Conference on Industrial Furnaces and Boilers, Porto, Portugal, 2006. (11) Arvelakis, S.; Folkedahl, B.; Dam-Johansen, K.; Hurley, J. Energy Fuels 2006, 20 (3), 1329. (12) Arvelakis, S.; Frandsen, F. J.; Dam-Johansen, K. Proceedings of the 2nd World Biomass Conference, 2004; 1427. (13) Arvelakis, S.; Frandsen, F. J. Biomass Bioenergy, in press.
10.1021/ef800097a CCC: $40.75 2008 American Chemical Society Published on Web 08/12/2008
Ashes from Energy Production and MSW Handling
Energy & Fuels, Vol. 22, No. 5, 2008 2949
Table 1. ICP-OES Elemental Analysis of the Co-combustion Fuels and Ash Samples, % (w/w)
Table 2. ICP-OES Elemental Analysis of the Economizer Fly Ash Samples, % (w/w)
brown bituminous OP18 OP19 OP19 OP20 coal RBF coal straw cyclone cyclone filter filter Al2O3 SiO2 K2O Na2O CaO MgO Fe2O3 P2O5 SO3 TiO2 Cl a
4.29 6.99 0.63 2.73 37.95 15.51 14.52 0.03 15.28 0.23 0.46
17.1 70.2 1.00 0.3 1.3 0.5 6.7 nda 1.00 0.7 nda
0.5 59.6 12.0 0.8 12.5 2.1 0.4 3.5 2.1 0.1 3.88
0.18 1.2 25.3 14.9 11.2 0.8 0.0 0.68 40.5 0.0 4.9
0.43 3.4 32.9 9.8 6.5 0.4 0.7 0.97 35.7 0.01 7.2
4.18 5.8 30.12 30.8 19.24 21.37 5.25 2.85 9.53 2.76 3.38 2.85 5.25 3.2 3.92 2.69 11.22 9.8 0.26 0.58 7.58 14.5
nd is not determined.
Management for Industrial Biomass Co-Combustion (OPTEB) (1996-1998) under contract JOR3-CT95-0057. A detailed description of the specific pilot plant is given elsewhere.14 The ash samples were collected from the cyclone and filter sections of the boiler. The four MSW ash samples studied here came from the Horsens incineration plant boilers 1 and 2 in Denmark. The ash samples were sampled from the economizer and the filter sections of the two boilers. A leaching pretreatment was applied to the MSW ash samples to remove the salt material from them before the viscosity measurements were performed. The leaching treatment consisted of the washing of the ash material using deionized water to remove all of the water-soluble salts present on the surface of the ash particles. In a second stage, the leached ash samples were used for viscosity measurements. This was performed to protect the hightemperature furnace used for the premelting of the ash samples before the viscosity measurements from the alkali metal, chlorineinduced corrosion. The removal of the salt material from the ash samples before the viscosity measurements was believed not to have a significant effect on the melting behavior of the ash samples, because the samples were heated first at temperatures over 1200 °C, where it is known that the alkali salts are vaporized to the gas phase. However, later, detailed investigation by the authors showed that the salts should not be removed before the viscosity measurements to avoid affecting the results produced.12,14 This pretreatment was not applied to the co-combustion ash samples. Table 1 presents the results from the inductively coupled plasma-optical emission spectrometry (ICP-OES) elemental analysis of the coal and wheat straw samples used to form the cocombustion mixtures and the ash fractions collected from the cyclone and filter sections of the PF boiler and used in the viscosity measurements. The cyclone (OP18 and OP19) ash samples had similar compositions. They were rich in alkali metals, sulfur, and chlorine and were average in calcium, while the concentrations of the other inorganic constituents were very low. OP19 had the highest content of alkali metals, sulfur, and chlorine. This was attributed to the higher proportion of straw into the specific fuel mixture compared to the other co-combustion mixtures. The filter ash (OP19 and OP20) fractions were rich in alkali metals, sulfur, and, especially, chlorine. However, the concentration of chlorine was higher now in the case of the OP20 filter sample, while the amount of straw was lower in the co-combustion mixture. This might be due to the different type of coal used in the co-combustion mixtures during the two tests and/or because of the different temperature conditions prevailing in the filter section during the tests. The silica content increased (30% w/w) because of the high silica content of the ash from the bituminous coal used in the cocombustion mixture. The very low temperature in the filter section led to a substantial condensation of alkali and chlorine vapors that had not been condensed previously in the cyclone section and the enrichment of the fly ash particles, especially, in chlorine. The (14) Arvelakis, S.; Jensen, P. A.; Dam-Johansen, K. Energy Fuels 2004, 18 (4), 1066–1076.
eco ash eco eco ash eco non-L 1 ash L 1 % difference non-L 2 ash L 2 % difference Al2O3 SiO2 K2 O Na2O CaO MgO Fe2O3 P2O5 SO3 TiO2 ZnO Cl
11.14 23.79 3.25 3.50 23.94 2.67 3.86 4.58 7.50 2.33 0.69 2.70
13.22 28.93 1.45 2.02 26.32 3.17 4.14 5.50 6.50 2.67 0.82 0.12
18.64 21.62 -55.56 -42.31 9.94 18.75 7.41 20.00 -13.33 14.29 20.00 -95.56
11.33 24.43 3.49 3.64 23.66 2.67 3.86 4.58 8.25 2.17 0.70 2.70
13.22 27.64 1.45 1.89 25.90 3.00 4.57 5.50 7.00 2.67 0.86 0.11
16.67 13.16 -58.62 -48.15 9.47 12.50 18.52 20.00 -15.15 23.08 23.21 -95.93
Table 3. ICP-OES Elemental Analysis of the Filter Fly Ash Samples, % (w/w)
Al2O3 SiO2 K2O Na2O CaO MgO Fe2O3 P2O5 SO3 TiO2 ZnO Cl
filter ash non-L 1
filter ash L1
% difference
filter ash non-L 2
filter ash L 2
% difference
8.69 15.43 9.04 7.01 21.70 2.33 2.14 4.31 10.00 1.67 2.12 12.70
12.28 23.57 1.21 1.62 29.96 3.67 3.14 6.41 9.75 2.33 3.12 0.08
41.30 52.78 -86.67 -76.92 38.06 57.14 46.67 48.94 -2.50 40.00 47.06 -99.40
8.69 15.64 9.04 7.14 21.84 2.50 2.00 4.31 10.00 1.67 1.99 13.10
11.71 23.36 1.14 1.62 30.10 3.67 3.29 6.41 10.00 2.33 2.99 0.04
34.78 49.32 -87.33 -77.36 37.82 46.67 64.29 48.94 0.00 40.00 50.00 -99.69
composition of the ash fractions was consistent with the composition of the individual fuels (coal and straw) used to form the cocombustion mixtures. Tables 2 and 3 present the results from the ICP-OES elemental analysis of the MSW ash fractions. As seen from these results, the different ash fractions sampled from boilers 1 and 2 had similar compositions. This is an indication that the firing conditions in the two boilers as well as the MSW feeding streams were similar. The ash fractions from the economizer area contained higher amounts of silica, aluminum, iron, magnesium, and calcium before the leaching process and lower amounts of alkali metals, zinc, chlorine, and sulfur compared to the fractions from the filter area. This is attributed to the higher temperatures prevailing in the economizer area compared to the filter area and resulted in a lower condensation of alkali salt vapors on the surface of the ash particles. Leaching led to a substantial extraction of alkali metals and chlorine from the different ash fractions. The highest mass loss was observed in the case of the filter ash fractions that also contained the highest amounts of water-soluble salts. Two different setups in the United States and Denmark were both purchased by the same company, HAAKE, but different versions of the basic instrument were used in this study to determine the reproducibility of the method. The high-temperature viscometer depicted in Figure 1a was used for the viscosity tests in Denmark at the Danish Technical University, CHEC Research Center. The viscometer consisted of two parts: the RV20 rotational viscometer and the ME1700 high-temperature furnace. During the measurement, the liquid was contained in an outer cylinder, a cup. A concentric inner cylinder, a spindle, was rotated at a steady speed in the liquid. The ME 1700 furnace is basically a vertical tube furnace with a maximum temperature of 1700 °C. The furnace is equipped with a protective inner ceramic tube that enables the establishment of an inert atmosphere. The viscometer is computercontrolled, and the viscosity data were collected every 30 s and put directly into a viscosity versus temperature plot. The viscosity measurements in the United States at the Energy and Environmental Research Center (EERC) at the University of North Dakota were performed using the HAAKE high-temperature rotational viscom-
2950 Energy & Fuels, Vol. 22, No. 5, 2008
ArVelakis et al. Table 4. Thermal Behavior of Ash Melts During the Various Stages of the Viscosity Measurements problems problems melting during the during temperature final ash premelting the experimental (°C) condition stage stage OP18 cyclone OP19 cyclone OP19 filter OP20 filter eco ash leached, Horsens 1 eco ash leached, Horsens 2 filter ash leached, Horsens 1 filter ash leached, Horsens 2
1400 1400 1400 1400 1300
silicate silicate glass glass glass
none none none none reaction
none none none none none
1300
glass
reaction
none
1350
silicate
reaction
ash spillover
1350
silicate
reaction
none
porcelain dish to burn any residual carbon that might have been left on the sample from the ashing process and then premelted in an 80:20 Pt/Rh dish using a muffle furnace for at least 24 h. During the premelting process, the melting temperature of the slag material and, subsequently, the working temperature of the viscometer were determined. After the premelting stage, the material was cooled, crushed with a hammer, and used to fill the crucible used in the viscosity measurements. Table 4 presents the results from the premelting process of the co-combusted ash fractions as well as the leached MSW ash fractions used for the viscosity measurements. The co-combustion ash samples with a low content of aluminosilicates, such as the cyclone OP18 and OP19 as well as the OP19 and OP20 filter ash, melted at rather low temperatures compared to the average melting temperature of the coal ashes, which is close to 1600 °C. The very low content of aluminosilicates did not allow for the formation of a glass during the melting of the ash and the subsequent cooling during the premelting or during the viscosity tests as seen for the OP18 and OP19 cyclone ash samples. A nonglassy deposit was produced. The samples with medium to high aluminosilicate contents produced a glassy slag during both the premelting and viscosity tests. In a similar way, in the case of the MSW ash samples, the higher amounts of silica of the economizer ash samples favored the formation of a glass during the premelting process, while in the case of the filter ash samples, a nonglassy silicate was formed. The MSW ash materials were very reactive during the premelting stage, where they appeared to react with the platinum/rhodium crucible used in the process mainly because of the presence of sulfur in the form of sulfides, as the metallographic examination of the crucible showed. The reaction resulted in the complete destruction of the crucible. During the viscosity measurements, only the MSW filter ash sample from boiler 1 showed problematic behavior, resulting in ash spillover from the crucible.
Results and Discussion Figure 1. Viscometer (a) setup in Denmark, (b) setup in the United States, and (c) setup in parts of the United States.
eter seen in Figure 1b. In this case, the crucible stands on the top of a cylindrical alumina pedestal seen in Figure 1c, with a diameter of 38.6 mm and length of 152.4 mm, during the measurement. During the viscosity measurement, the slag is heated to a temperature 50 °C higher than its liquid temperature and then the viscometer bob is submerged into the slag until the slag just covers its top, at which point it is rotated at 45.3 rpm. The viscometer, in this case, was manually controlled, and therefore, measurements of the viscosity were taken at temperature segments of 50 °C as the ash melt cooled from its liquid temperature. The ash melt is left to equilibrate for 30 min in each new temperature before the viscosity is measured. In both cases, before the start of the measurement, the ash material was preheated to 900 °C in a
Co-combustion Ash Samples: Cyclone Ash Samples. Figure 2 presents the results from the viscosity runs using the cyclone (OP18 and OP19) ash samples. In all figures, the continuous lines present the results from the viscosity measurements performed in Denmark, while the dotted lines display the tests performed in the United States. The viscosity curves, in the case of the measurements performed in Denmark tests 1 and 2, using rotational speeds of 10 000 and 5000 rpm, respectively, for the OP18 cyclone sample showed that this material displayed Newtonian flow behavior in the first steps of the measurement from 1360 °C down to 1280 °C where the viscosity of the melt changed almost linearly with the temperature, independent of the shear rate. However, the viscosity values partially fluctuated in the same temperature segment, indicating the presence of solid phases to an extent
Ashes from Energy Production and MSW Handling
Figure 2. Viscosity tests with the biomass cyclone ash samples: (a) OP18 cyclone ash and (b) OP19 cyclone ash.
into the melt. The behavior of the ash melt changed to nonNewtonian below 1280 °C, where a sudden and large increase of the viscosity was observed. The ash melt there reached its temperature of critical viscosity (Tcv) point, with the amount of solids in the melt increasing significantly, leading to the solidification of the melt at 1246 °C for test 1 and at 1261 °C for test 2. Repetitions of the measurement performed using a lower rotational speed of 3000 rpm below 1300 °C showed that the presence of solid phases in the ash melt was permanent at this temperature level, which led to higher viscosity values measured. The non-Newtonian behavior of the melt at this temperature segment did not allow for the reproduction of a measurement using the same rotational speed, although the shape of the produced curves was similar. The results from the tests performed in the United States showed that the ash melt followed Newtonian behavior in a larger temperature segment compared to the measurements performed in Denmark. The melt reached its Tcv point in the temperature range of 1240-1265 °C. The viscosity curves from the measurements in the United States showed lower viscosity values but in the same range compared to those of the measurements performed in Denmark. The chemical composition of the OP18 cyclone ash showed that the ash contained large amounts of alkali metals, sulfur, calcium, and chlorine and small amounts of aluminosilicates, magnesium, etc. The produced melt had a similar composition with higher calcium and sulfur values and lower alkali metals and chlorine because of partial evaporation and decomposition at this temperature level of many of the alkali- and chlorinecontaining phases.14 As a result, the ash material is considered to be only partially molten and contain a rather large number of solid phases, such as CaO and MgO, as well as decomposing phases, such as CaSO4, and possibly solid oxides, such as mixtures of Fe2O3/K2O and/or Fe2O3/Na2O, which led to high
Energy & Fuels, Vol. 22, No. 5, 2008 2951
viscosity values and non-Newtonian behavior during the viscosity measurements in the specific temperature segment (1400-1200 °C). The 30 min equilibrium time applied in the case of the viscometer in the United States before a viscosity value was taken seems to have had a positive effect on the determination of the melt viscosity, resulting in a larger temperature segment where the melt behaved as a Newtonian fluid. It is possible that the equilibrium time allows for the partially molten sample to homogenize better and, thus, to show Newtonian behavior in a larger temperature segment now compared to the measurements in Denmark. The results from the viscosity measurements with the OP19 cyclone ash sample showed different viscosity behavior than the OP18 cyclone. Figure 2b shows that the melt followed Newtonian behavior in the first steps of the viscosity measurement (test 1). The specific measurement terminated abruptly at 1176 °C because of a software failure. Repetition of the measurement under exactly the same conditions resulted in a good reproduction of the initial measurement. The melt showed Newtonian behavior until the temperature of 1110 °C. The Tcv point appeared at 1109 °C, where the melt solidified rapidly. Several measurements were also performed, starting from lower temperatures and with lower rotational speeds (5000-10 000 rpm). These measurements are divided into two groups. The first group included tests 4, 5, and 6. These tests reproduced well the viscosity curves shown in tests 1 and 2, and test 6 produced the viscosity curve that showed Newtonian behavior for temperatures as low as 1076 °C with Tcv and the solidification of the melt starting at 1069 °C. The second group includes test 3. This test was started from a lower temperature and with a rotational speed of 10 000 rpm. The viscosity curve is seen to take higher values in this case and to be very different compared to the other viscosity curves. The Newtonian flow behavior is seen to take place in the temperature segment of 1250-1139 °C, and the Tcv point is seen at the 1139 °C. The viscosity measurements performed in the United States showed the same melt behavior compared to those performed in Denmark. Both tests were seen to have a very good match, while the viscosity curves presented in Figure 2a were seen to follow the same pattern and to be very close to those performed in Denmark. The ash melt reached its Tcv point at higher temperatures (1148-1173 °C) compared to the measurements performed in Denmark. The chemical analysis of the OP19 cyclone ash showed that the ash was rich in alkali metals, sulfur, and chlorine and contained average amounts of calcium and low amounts of the other elements. The lower amounts of calcium are believed to be responsible for the improved Newtonian behavior shown by the ash sample compared to the OP18 cyclone. The lower amount of calcium in the melt resulted in lower amounts of solids, such as CaO and CaSO4, and, subsequently, in lower viscosity values as well as larger Newtonian flow behavior. The lowest temperature where the melt was seen to solidify was at 1069 °C, which corresponds to the melting point for K2SO4, which is believed to be the main compound in the melt.13 Filter Ash Samples. Parts a and b of Figure 3 present the results from the viscosity tests using the OP19 and OP20 filter ash samples using the two viscosity setups in Denmark and the United States. As seen in Figure 3a, the viscosity measurements correlated well in all cases in the temperature segment where the melt behaved as a Newtonian fluid. In the measurements performed in Denmark, the ash sample melted at 1260 °C and the Tcv was present in the temperature range of 1149-1134 °C, depending upon the different measurements. After the Tcv point,
2952 Energy & Fuels, Vol. 22, No. 5, 2008
ArVelakis et al.
Figure 3. Viscosity tests with the biomass filter ash samples: (a) OP19 filter ash and (b) OP20 filter ash.
the melt behaved differently in every viscosity measurement. The melt showed different solidification points ranging from 1149 to 1095 °C. This showed that the specific melt behaved as a multiphase mixture in temperatures below Tcv, with solid and molten phases forming and dissolving periodically that also affected the viscosity behavior as well as the solidification point. The measurements performed in the United States started at a higher temperature (1350 °C versus 1260 °C) compared to those performed in Denmark. The measurements showed that the ash melt behaved as a Newtonian fluid in the same temperature segment where Newtonian behavior was also seen in the case of the tests performed in Denmark. The Tcv point of the melt was now seen to be in the same area as in the case of the measurements using the viscosity setup in Denmark. However, the solidification of the sample proceeded faster and the melt solidified at 1148 °C. Figure 3b presents the results from the viscosity measurements using the OP20 filter ash sample. The measurements performed in Denmark showed that the ash melt behaved as a Newtonian fluid in the temperature segment of 1360-1220 °C in all measurements. The Tcv was seen in most measurements starting at high temperature in the area of 1220-1207 °C. After that point, the melt showed non- Newtonian behavior, and the viscosity values depended upon the shear rate. The melt crystallized fast, and solidification occurred around 1190 °C. An exception was observed in the case of test 4, where the very low rotational speed (2500 rpm) used showed the ash melt to follow Newtonian behavior until 1150 °C. Tcv appeared at 1153 °C, and the melt solidified at 1131 °C, although the measurement starting temperature was now lower compared to the other tests. The measurements performed in the U.S. setup showed good agreement with the measurements performed in Denmark. Both U.S. tests, as presented in Figure 3b, followed Newtonian
Figure 4. Viscosity tests with the MSW economizer ash samples: (a) eco ash leached Horsens 1 and (b) eco ash leached Horsens 2.
behavior in the same temperature segment as the majority of the measurements performed in Denmark. The viscosity values in the case of the U.S. measurements were almost 1 order of magnitude lower compared to the measurements performed in Denmark, while in the case of the OP19 filter ash sample, the viscosity values matched almost completely. However, the observed difference is not considered to be significant if we consider that even the measurements in the same setup in Denmark present a deviation of around 10-50%, depending upon the starting temperature and the rotational speed. MSW Ash Samples: Economizer Ash Samples. Figure 4a presents the results from the viscosity tests using the economizer ash leached sample from the Horsens 1 boiler. In total, three viscosity tests using various rotational speeds varying from 1000 to 10 000 rpm were performed in Denmark and two tests using the same rotational speed of 45.3 rpm were performed in the United States. The results from the viscosity tests performed in Denmark showed that the ash melt showed Newtonian behavior in the beginning of the viscosity measurements in all cases. The tests started at 1300 °C generated curves that matched very well in all cases. The Tcv point appeared in the temperature segment of 1129-1101 °C in all cases. The results from the viscosity measurements performed in the United States show a very good match with the results from the measurements performed in Denmark. The viscosity values were now slightly higher in the beginning of both measurements around 1300 °C compared to those from the measurements performed in Denmark. The ash
Ashes from Energy Production and MSW Handling
melt again showed Newtonian behavior in the temperature segment of 1300-1121 °C, which almost matched the temperature segment for Newtonian behavior for the measurements performed in Denmark. The transition to non-Newtonian behavior started at 1121 and 1105 °C, respectively, for the two measurements. After the Tcv point, the melts solidified fast, as in the case of the viscosity tests performed in Denmark. The slightly higher viscosity values in this case are attributed to the significantly lower rotational speed compared to the tests performed in Denmark. Figure 4b presents the results from the viscosity tests with the economizer ash leached sample from the Horsens 2 boiler. The results from the viscosity tests performed in Denmark show that this ash melt followed Newtonian behavior in the temperature segment of 1300-1113 °C. The viscosity tests from the different test runs matched very well. The melt showed its Tcv at 1113 and 1109 °C, respectively. However, the test 1 shows a Tcv and a solidification point at significantly higher temperatures compared to the other tests. The rotational speed used in all tests was 5000 rpm. This provides an indication that the different Tcv and solidification point in the case of test 1 is probably caused because of the presence of a large solid phase in contact with the spindle, which is not reappearing again in the other tests. The viscosity test 4 that was started at a temperature below 1200 °C did not correlate well with the tests started at higher temperatures. This test produced a curve where the melt shows lower viscosity values compared to the tests started at higher temperatures and higher rotational speeds. The results from the viscosity tests performed in the United States with the second type of viscometer show a good match with the results from the tests in Denmark. The melt followed Newtonian behavior in the temperature segment of 1330-1108 °C, while the viscosity values were slightly higher compared to the viscosity from the tests in Denmark. This is probably due to the substantially lower rotational speed used now that might favor the solidification of the melt. A comparison of the results with the results from the viscosity tests using the previous ash sample shows that there is a good agreement in the case of the tests performed in Denmark as well as in the case of the tests performed in the United States. This is expected, because the two ash samples had similar compositions, as seen in Table 2. However, the viscosity curves from the tests in the United States showed a better match than the curves from the tests in Denmark. Filter Ash Samples. Figure 5a presents the results from the viscosity measurements performed using the filter ash sample from the Horsens 1 boiler using the two different viscometers. The filter ash is shown to have melted in the same temperature range as the economizer ash. However, the ash melt solidified at a higher temperature, and it followed Newtonian behavior in a shorter temperature segment compared to the economizer ash melt. The viscosity curves from the measurements in Denmark show that the filter ash displayed Newtonian behavior in the temperature segment of 1340-1227 °C, while the curves from the different measurements were in close proximity but showed small to medium deviations during the measurements, probably because of a substantial non-uniform solids/crystal formation. The Tcv point in all cases was in the range of 1227-1202 °C, and after this point, the melt solidified rapidly. The results from the viscosity measurements performed in the United States are in good agreement with the results from the measurements in Denmark. As seen in Figure 5a, the viscosity values from the tests in the United States are very close to those from the tests performed in Denmark. Below 1250 °C, the viscosity
Energy & Fuels, Vol. 22, No. 5, 2008 2953
Figure 5. Viscosity tests with the MSW filter ash samples: (a) filter ash leached Horsens 1 and (b) filter ash leached Horsens 2.
curves from the tests in the United States showed lower viscosity values but were still very close to those viscosity values of the tests performed in Denmark. The curves showed some deviations during the Newtonian flow behavior of the ash melt, probably because of amorphous solid phases/crystal formation, as in the case of the tests in Denmark. The Tcv point in this case is seen to be at 1207 °C, which is very close to the lowest Tcv point observed for the tests in Denmark (1202 °C). The ash chemical composition presented in Table 3 is in good agreement with the results from the filter ash viscosity measurements. The filter ash contained large amounts of calcium and elements, such as phosphorus and sulfur, as well as low amounts of silica and aluminum. This composition does not favor the formation of a constant silicate network, which was also seen in the premelting stage, but rather the formation of a multiphase melt where the different phases solidified or crystallized separately and created deviations from the Newtonian flow behavior and a higher Tcv point during the cooling of the melt. Figure 5b presents the results from the viscosity measurements using the filter ash from the Horsens 2 boiler. The results from this ash melt are different from those of the previous filter ash melt. In the case of the measurements performed in Denmark, the ash melt followed Newtonian behavior and showed a Tcv point in most cases close to 1200 °C, as in the case of the previous ash melt. However, the results from measurement 1 are different. The ash melt followed Newtonian behavior until the temperature of 1163 °C was reached, at which point a sudden decrease of viscosity was observed. The viscosity values increased again below 1130 °C, and the melt was seen to reach its solidification point at 1095 °C. The rotational speed of this test is 5000 rpm, which is between the rotational speeds used in the other two tests (3000 and 10 000 rpm). This gives an indication that the rotational speed can affect the solidification behavior of low-Si melts that tend to solidify as non-uniform
2954 Energy & Fuels, Vol. 22, No. 5, 2008
multiphase solutions. The results from the tests in the United States followed the same pattern. The ash melt in both cases followed a Newtonian flow behavior until the temperature of 1151 °C was reached, at which point the same large drop in viscosity was observed. The viscosity of the melt increased again below 1110 °C, and the melt solidified at 1028 °C. This behavior was attributed to the formation of thixotropic phases into the melt. These phases have the tendency to dissolve easily when in contact with the spindle and to lower the viscosity substantially. These results comply well with the ash elemental analysis and provide proof that, in low silica ashes, the formation of multiphase melts is the main route through which these ash melts solidify.4,12,13 Conclusions The results from the viscosity measurements show very good agreement in most cases between the two viscometers. The U.S. noncomputer-controlled viscometer showed the smallest deviations in the viscosity measurements. This is attributed to the rather large time needed for data acquisition (30 min) compared to the computer-controlled viscometer (30 s). The computercontrolled viscometer had higher sensitivity regarding the detection of viscosity changes because of melt composition changes. However, the computer-controlled operation mode often led to false results, indicating complete solidification of the melt when the viscometer spindle was in contact with a rather large solid-phase and/or crystal. This results from the operating mode of the computer-controlled viscometer. Because
ArVelakis et al.
the crucible used for the viscosity measurements is now mounted into the alumina pedestal, the only way to avoid the destruction of the spindle in the case of a 100% torque signal is to automatically stop the measurement. This is mainly observed in the viscosity measurements starting at lower T compared to the original ash melting point. The computer-controlled viscometer is considered to be more suitable for the determination of viscosities from Si-rich ash melts that tend to solidify in a uniform way, producing a constant silicate network. The viscosities of ash melts low in silica and rich in calcium and other alkali, alkaline-earth metals, and compounds such as iron and phosphorus are better determined using the noncomputer-controlled viscometer. These melts form multiphase solutions, where solid and liquid phases come into contact with the viscometer spindle in a completely random way, and as a result, a larger data acquisition time is needed to produce more accurate viscosity values. In the case of the low-Si multicomponent melts, the viscometer rotational speed had a direct effect on the viscosity curves as seen in the case of the filter ash from the Horsens 2 boiler. Acknowledgment. This work is part of the Combustion and Harmful Emission Control (CHEC) research program funded by the Technical University of Denmark, Elsam A/S, Energy E2 A/S, PSO funds from Eltra A/S and Elkraft A/S, and the Danish Energy Research Program. EF800097A