292
Ind. Eng.
Chem. Fundam. 1983, 22, 292-298
Rasaiah, J. C. J . Chem. Phys. 1970, 5 2 , 704. Rasaiah, J. C. J . Chem. Phys. 1972, 56, 3071. Robinson, R. A.; Stokes, R . H. "Electrolyte Solutions", 2nd ed.; Butterworths. London, 1959. TiTien. H. J . Phvs. Chem. 1983. 67. 532. Triolo, R.; Grigha, J. R.; Blum, L. J . Phys. Chem. 1976, 80, 1858. Triolo, R.; Blum, L.; Floriano, M. A. J . Phys. Chem. 1978, 82,1368. Triolo, R.; Blum, L.; Floriano, M. A. J . Chem. Phys. 1077, 6 7 , 5956. Valleau, J. P.; Card, D. N. J . Chem. Phys. 1972, 5 7 , 5457.
Waisman, E.; Lebowitz, J. L. J . Chem. Phys. 1070, 52. 4307. Waisman, E.; Lebowitz. J. L. J . Chem. Phys. 1972, 56, 3086. Wiechert, V. H.; Krienke, H.; Feistel, P.: Ebeling, W. 2.Phys. Chem. Leipzig 1978, 259, 1057.
Received for review September 14, 1981 Revised manuscript received December 28, 1982 Accepted January 19, 1983
Analysis of Sulfur and Nitrogen Pollutants in Three-phase Coal Combustion Effluent Samples Jeffrey R. Burklnshaw,'
L. Douglas Smoot, Paul 0. Hedman, and Angus U. Blackham
Combustion Laboratory, Brigham Young University, Provo, Utah 84602
Methods for analysis of sulfur and nitrogen pollutants sampled with water-quenched probes from pulverized coal combustors have been developed. Analyses are outlined for measurement of SO2, H2S, COS, CS2, and NO in the gas-phase sample, S O:-, S2-,CN-, and NH', in the liquid-phase sample, and total sulfur and nitrogen in the solid-phase sample. Permanent gas analysis and solid ultimate analysis methods are also presented. The relative error for these measurements was 6 to 8 % , based on extensive laboratory testing. These methods were applied to the analysis of three-phase (gas, solid, and liquid) samples from a laboratory-scale pulverized coal combustor in a series of three tests with two different coals. Results emphasized sulfur pollutant measurements. Sulfur oxides, and to a lesser extent, H2S, were the principal sulfur species formed while low levels of COS and CS2 were also detected. Most of the SO, and H2S were recovered in the liquid-phase sample. A mass balance for sulfur including that remaining in the char agreed to within 1% when the coal feed rate was determined from a carbon balance.
Introduction The processes that control within a coal combustor are highly speculative if the only test data available are the analyses of the initial reactants and the final products. Furthermore, optimization of the combustion process becomes almost totally empirical in nature. To provide greater insight and a sound theoretical foundation for combustion technology and modeling, local samples taken from within the reactor are essential. Such local samples may be obtained with a probe which water-quenches the reactants and products to provide a representative sample (Thurgood et al., 1980). The chemical analysis of the three-phase (gas, liquid, and solid) sample is then an essential endeavor. Price et al. (1983) reported a similar but less extensive work for the analysis of samples from a coal gasifier (see also Skinner et al., 1980). Several investigators have reported methods and results for the measurement of each of the pollutant species of interest in this study: SO2,H2S,COS, CS2, SO:-, S2-, CN-, NH,+, NO, N(solid), and S(so1id). Burkinshaw (1981) provides a summary of the independent work from 24 studies relating to experimental measurement of one or another of these pollutants. However, no studies have been reported wherein the determination of all of these species have been considered from a single sample. A strategy is outlined for the analysis of the three-phase sample obtained from a water-quenched probe in a laboratory-scale coal combustor. The relative error of each analytical method is established based on extensive laboratory testing (Burkinshaw, 1981). Experimental Facilities Reactor. The coal combustor is shown schematically in Figure 1. Coal is extrained in the central primary air
* Phillips Petroleum Co.. Bartlesville, OK. 0196-4313/83/1022-0292$01.50/0
stream and transported into the reactor where it mixes and reacts with the gas from the coaxial secondary air stream. The secondary air flow can be directed into the reactor parallel to the primary jet or can be given a swirling flow in order to control mixing rates and flame ignition. Local samples of combustion products (both gaseous and particulate) are removed from the reactor with a waterquenched sample probe. The reactor is of modular construction so that the probe section can be located at various axial positions. Then with a radial traversing mechanism on the probe, samples were obtained at various radial locations. Sample Probe. A schematic of the probe design is shown in Figure 2. Distilled water was passed through a closed manifold to the probe tip where it was sprayed into the sample duct, rapidly quenching the chemical reactions, cooling the char to a non-sticky form, and preventing char deposition on the probe walls. It is difficult to obtain a representative gas and char sample from a reactor with a sample probe, especially in a turbulent flow where large density gradients exist and where gases and particles may flow in different directions. However, an effort was made to sample isokinetically (i.e., with the gas velocity in the probe inlet the same as the gas velocity in flow-field just ahead of the probe). This was attempted by balancing the static pressure measured inside the probe tip (see Figure 2a) to the local pressure in the reactor. The sampling rate through the probe was varied until these two pressures were nearly balanced. Balancing was routinely accomplished in tests without swirl or when the probe was located near the bottom of the reactor. Balancing of sample probe pressures in regions of strongly swirling flow was more difficult. A V-shaped probe (Figure 2b) was used. This probe was rotated into a highly swirling flow without change in radial or axial position. Alignment of this V-shaped probe to the 'C 1983 American Chemical Society
Ind. Eng. Chem. Fundam.. Vol. 22, No. 3, 1983 293 Coal
Stream
Primary Air Stream
Moveable Block Swirl Generator ~
Preheat Methane
1.6 cm Wall Temperalure I18 iocationr)
sn,,,: Ch,, slll.e
Figure 3. Schematic diagram of sample system
Steel Caring
Ceramic Insulation
-~ . -
Figure 4. Gas analysis scheme Exhaust c-
+
Quench Water
!
r% Figure 1. Schematic of atmospheric combustor. 1.11 cm O.D. Stainless Steel Tubing
i L0.16 cm i.D. Static Pressure Tap 81 Probe
Tip
&F:-q? 9.7cm
p-48.3cm-~
2.54 cm
bl Swirl Probe
Figure 2. Schematic diagram of sample probe and probe tip.
swirling flow direction was accomplished by maximizing the gas sample flow through the probe after establishing near-isokinetic collection with the probe set near the correct flow angle. Sample System. A sample taken from the reactor consisted of gas, char, and an aqueous solution. Before
analysis, it was necessary to separate the three phases. This was done in the sampling system shown schematically in Figure 3. The sample first passed through a Teflon disengager where the gas was separated from the solution-char mixture. The gas was then cooled in an ice bath and passed through a second disengager to remove any additional liquid condensed in the ice bath. Next it passed through a fine particle filter and a mass-flow meter. The cleaned gas was either stored in a glass storage bottle for later gas chromatographic analysis or passed to the on-line gas analysis instruments. The solution-char sample was collected in a glass vessel, analyzed for dissolved gases (e.g., H2S,SO,, NH3, HCN), and filtered. The char was washed, dried, and analyzed for elements (C, H, 0, N, and S) and for ash. Gas Tracer. An inert gas tracer (argon) was introduced into the inlet primary air stream. This tracer permitted the calculation of the local gas mixture fraction and hence the mixing rate between the primary and secondary gas streams. The gas tracer also permitted the calculation of carbon humout from the measured gas composition of the combustion products, a forced argon mass balance, and the coal feed rate. The flow rate of the argon trace gas was high enough (i.e., about 5 mol% of primary stream) to he measured with the gas chromatograph, but not high enough to interfere significantly with the reaction process. Gas Analysis Procedures The overall scheme for gas phase analysis is illustrated in Figure 4. A discussion of analysis procedures for each species follows. Sulfur Gas Analysis. The gas stream from the sample collection system was diverted through a 125-mL glass sampling bulb fitted with a rubber septum. A gas-tight syringe was used to withdraw a 0.5-mL sample from the sampling bulb through the septum. The sample was then
Ind. Eng. Chem. Fundam., Vol. 22, No. 3, 1983
204 40
-
____ ---- 7
Collection Jar
0
/
m'
Aliquot
Shake with Ion Exchange ReWl Res n
Filter
100 ml
20
40
50
50
100
120
2 140 150
Peak Area mV sec
Figure 5. Examples of sulfur gas calibration curves.
Preserve wilh 1 mi NaOH
1 Precipitate S with PblN0312
1
I
50 IPI Alicuo'
Preserve with
1
1
Preserve and Prepare with 50 mi S Buffer
0-
100 ml Aliquot
t %Specific Electrode
1 7110
15 HCI
Make basic with 1 r l 10M YaOH
I
NH3 Specilic Electrode
I
1
S
IfvHjl
I
i
Oxidize with 5 mi 30'0 H 2 0 2
4 Take 10 mi Aliquot
,
1
Filter
Add 5 drops Sullonazo Ill Indicator and ' 0 ml acetone
I CN Specific Electrode
I
Titrate with 0 0015 BalCIOL)2
1 ICN
I
t
IS041
Figure 6. Liquid preservation and analysis scheme.
injected into a Hewlett-Packard 5732A chromatograph equipped with a flame photometric detector (FPD). A 1.8 m X 1/3 cm Teflon (FEP) Chromasil310 column was used to measure SO2, H2S, COS, and CS2 in the range of 1 to 200 ppm. Calibration curves were constructed immediately prior to analysis, by use of gas standards for each sulfur species, as shown in Figure 5 . A pretreatment of the gas sampling bulb was necessary to prevent adsorption of the sulfur gases on the walls, particularly the small amounts of COS and CS2(Gangwall and Jayanty, 1979). The sampling bulb was filled with 50 ppm of SO2 in N2 several hours before the sampling time. At least 8 h later, the bulb was flushed with air at a flow rate of at least 1L/min for 10 min, after which the bulb was closed. This procedure covered active sites which would have otherwise adsorbed the sample sulfur gas species. At least seven or eight volumes of sample gas were required to purge the sample bulb and thus provide an undiluted and uncontaminated sample. Permanent Gas Analysis. Nitrogen (N2),methane (CH,), carbon monoxide (CO),and carbon dioxide (CO,) were measured using an HP-5832A chromatograph equipped with a thermal conductivity detector (TCD). Oxygen and argon (Oz-Ar) were measured collectively with this chromatograph because of the difficulty in separating these two gases. Peak areas were integrated and proportionality constants were calculated by an HP-18850A GC microprocessor terminal. Hydrogen was used as the carrier gas and the columns used were a 1.8 m X 1/3 cm stainless steel Porapak Q 80/100 and a 5 m X 1/3 cm stainless steel Molecular Seive 5A 60/80. The chromatograph was standardized with a laboratory gas standard consisting of 4.68% 0 2 , 1.86% Ar, 75.84% N2, 1.86% CHI, 3.75% CO, 11.07% C02, and 0.94% H2 (by volume). The sample gas was injected into the chromatograph by water displacement of a 25-mL sample through the 0.5-mL loop of the gas sampling valve (GSV). Argon, Hydrogen, and Oxygen Analysis. Both argon and hydrogen were measured using a Tracor MT-150 gas chromatograph equipped with an ultrasonic detector (USD).Peak areas were integrated with an HP-18850A GC microprocessor terminal. Oxygen was used as the carrier gas, thus allowing the direct measurement of argon. The column was a 5 m X 1 / 3 cm stainless steel Molecular Sieve 13X 60/80. The chromatograph was standardized with the gas used in permanent gas analysis. The sample gas was injected into the chromatograph by water displacement using a 0.5-mL GSV as in the permanent gas analysis. With collective Ar-O2 chromatographic measurement (H2carrier) and the independent argon chromatographic measurement (0, carrier), gas concentrations of argon and
oxygen were computed, accounting for differences in thermal conductivity of the two species. Nitric Oxide Analysis. Nitric oxide in the combustion sample was analyzed with a Thermo Electron Model 10 AR chemiluminescent analyzer. In the detector, ozone reacts with NO to form NOz in an excited state which emits light. The intensity of the light emitted is monitored by a photomultiplier tube located behind the appropriate optical filters to provide a signal that is proportional to the amount of NO present. The analyzer was incorporated into the system for online analysis by Harding et al. (1982) as illustrated in Figure 3. The instrument was calibrated with a standard gas (219 ppm NO in nitrogen) prior to the experiment. Rees et al. (1980) have shown that the response was linear over the range of NO levels observed and that it was sufficient to use a single gas standard for daily calibration. Liquid Sampling and Analytical Procedures The liquid solution and solid samples, which were separated from the gas sample in the gas/solution-char disengager, were collected in 1.5-L glass vessels. The collection rate was approximately 300 mL/min. Figure 6 shows a schematic representation of the solution preservations and the analytical procedures, while a detailed discussion of these procedures follows. Sulfide Analysis. Sulfide (S2-)was measured using an Orion Model 94-16 Ag+/S2- specific electrode in conjunction with an Orion Model 90-02 double junction reference electrode (Baumann, 1974). An Orion Model 701A digital electrometer with an Orion Model 605 electrode switch was used to display the relative potential. Standard sulfide solutions were prepared by dissolving sodium sulfide (Na2S)in distilled water. The concentration of the saturated S2- solution was determined by potentiometric titration with Pb(N03)2or by oxidation of S2-to SO$-and titration with Ba(C104)zin the presence of Sulfonazo I11 indicator (as in the sulfate analysis below). Calibration curves were constructed weekly by preparing a series of five S2-solutions in the concentration range to be studied. The relative potential in millivolts was determined as a function of the natural logarithm of the S2concentration. Figure 7 illustrates a S2-calibration curve and the deviations obtained for equivalent standard solutions measured on each of five consecutive days. For sample analysis, a 50-mL aliquot of the sample solution was added to a 100-mLpolypropylene storage bottle. A 50-mL aliquot of a specially prepared sulfide buffer was added to the sample to prevent oxidation of the S2-. The storage bottle was capped tightly and the solution was cooled in an ice bath. When cool, the solution was poured into a 250-mL beaker and the relative potential was
Ind. Eng. Chem. Fundam., Vol. 22, No. 3, 1983
-[-I
Perkin Elmer Model 240 Elemental Analyzer
Solid Sample
295
i: 2 E Wt O h H Wt 9 s A h; Wt 90 0
\ Sample
'Oxygen weight percent
IS
Sullur Analyzer
determined by difference
Figure 8. Solid analysis scheme. Relative Potential,
s=
:,
I
mV I
I
-770
-780
-750
-760
CN-: t
I
-160
-150
NH3: I
-140
-130
-120 I
t
10
20
30
40
50
60
Figure 7. S2-, CN-, and NH3 specific electrode calibration curves with standard deviation ranges measured over a 5-day period.
measured with the S2-specific electrode. Sulfate Analysis. Sulfate (S042-)was measured by titration with a 0.001 M Ba(C104)2solution. A 0.1% Sulfonazo I11 solution was used as an indicator (Budesinsky, 1975). In the absence of Ba2+ions, the indicator was red-violet, but in the presence of Ba2+it turned blue-green. Acetone was also present during the titration to increase the sensitivity of the indicator. The S042-solution was standardized against solutions of analytical grade potassium sulfate (K2S04). A 50-mL aliquot of the sample was transferred from the liquid collection jar to a 100-mL polypropylene bottle for storage. Prior to analysis, the solution was shaken for 10 min with a VWR Model 58815 vortex mixer with 1 g of Dowex 5OW-8X ion-exchange resin to remove any heavy metal ions which interfere with the titration. The solution was then filtered using vacuum fitration. A 0.8-pm mixed cellulose acetate and nitrate filter was used. Approximately 5 mL of 30% H202was added to oxidize any SO$to SO4". A 10-mL aliquot of the solution along with 10 mL of reagent grade acetone and 5 drops of 0.1% Sulfonazo I11 solution was then titrated with 0.001 M Ba(C104)2 titrant from red-violet to a blue-green end point. Any SO3 originally present in the gas sample was also detected by this aqueous-phase titration. Thus, since SO3 is highly soluble, the reported SO2 measurements contain SO3 as well and is often referred herein to as SO,. Cyanide Analysis. Cyanide (CN-) was measured using an Orion Model 94-06 CN- specific electrode with an Orion Model 90-02 double junction reference electrode. The relative potential was monitored on an Orion Model 701A digital electrometer. Standard cyanide solutions were prepared using reagent grade potassium cyanide (KCN). A weekly calibration curve was prepared from a series of five CN- solutions. Figure 7 shows a calibration curve with the deviations from that curve measured on the five consecutive days following the original calibration. The relative potential vs. the natural logarithm of the CNconcentration was plotted, yielding a linear calibration fit. The CN- concentration was measured by transferring a 100-mL aliquot of the sample to a 100-mL polypropylene sample bottle. Then, 1mL of 10 M NaOH was added to preserve the CN-. The bottle was then capped until it was analyzed. Since sulfide reacts with the cyanide specific electrode, impairing ita function and shortening the electrode life, any S2-was removed before analysis. A 1-mL aliquot of 0.1 M Pb(N03)2was added to the solution. The
formation of a black precipitate (PbS) indicated S2-was present and the precipitate was filtered out before analysis. The solution was then measured with the cyanide specific electrode (Luthy and Bruce, 1978). Ammonia Analysis. Ammonia (NH,) was measured with an Orion Model 95-10 NH3 gas-sensing electrode. This electrode was self-referencing and required no auxiliary electrode. The relative potential of this electrode was monitored on the Orion Model 701A digital electrometer. Although this electrode measured NH,, this species actually existed as NH4+prior to analysis. The NH4+was converted to NH3 immediately before analysis by raising the sample solution from a pH of about 5 to a pH of about 10. Standard solutions of NH3 were prepared using reagent grade ammonium chloride (NH4C1). A weekly calibration curve was prepared from a series of five NH3 solutions. Figure 7 also shows the deviations from the calibration curve for NH3 measured on the five consecutive days following the original calibration. As with the other ion-specific electrodes, the relative potential vs. the natural logarithm of the NH3concentration was linear. For sample analysis, a 100-mL aliquot of the liquid was placed in a 1WmL polypropylene sample bottle along with 1 mL of 0.1 M HC1 which lowered the pH to about 5 to preserve the species as NH4+ and thus prevent its volatilization and loss as NH3 gas. Immediately prior to analysis, 1mL of 10 M NaOH was added to the solution converting NH4+ to NH3, the species measured by the electrode. The electrode was immediately placed in the solution and the relative potential measured. Solid Sampling and Analysis Analytical procedures for the solid residue are illustrated in Figure 8. After the analysis, all liquid samples were returned to the collection jar. The solids were then separated from the liquid using vacuum filtration. A 0.8-pm mixed cellulose acetate and nitrate Millipore filter was used. The solid was then transferred to an aluminum drying dish and stored in a desiccator. The moisture was driven off from the solid sample by placing it in an oven at 104-110 "C for 1h (ASTM Method D 3173,1976). After drying, the sample was stored in an air-tight polyethylene vial for later analysis. Sulfur Analysis. A LECO Model 634-800 sulfur analyzer was used to determine the weight percent of sulfur in the solid sample. The solid sulfur was combusted to SOzin the analyzer. The SO2 was dissolved in an aqueous solution and the sulfite was titrated automatically with a standardized solution of potassium iodate (KI03) in the presence of starch indicator and potassium iodide (KI). The analyzer was standardized using 0.29%, 1.01% or 2.02% sulfur-in-hydrocarbon calibration standards. The weight percent of sulfur was measured by weighing approximately 50 mg of the sample exactly in a ceramic combustion crucible along with approximately 0.5 g copper and 2 g magnesium oxide (MgO). A porous lid was placed
296
Ind. Eng. Chem. Fundam., Vol. 22, No. 3, 1983
on the cricible and it was combusted in the analyzer induction furnace in an oxygen stream of about 1 L/min. The SO2product was bubbled through the titration solution and sufficient iodate (IO$-) titrant was automatically added to restore the I2 lost by reaction with S032-. The weight percent sulfur was then determined from the amount of titrant used. Elemental Analysis. The weight percents of carbon, hydrogen, nitrogen, ash, and oxygen in a sample were measured with a Perkin-Elmer Model 240 elemental analyzer. Reagent grade acetanilide (C6H5NHCOCH3)was used as the primary standard to obtain the necessary proportionality constants. Approximately 1mg of the sample was weighted exactly in a platinum boat. The boat was transferred to a combustion ladle and the ladle was inserted into the furnace. The sample was combusted with ultrapure oxygen (99.99% minimum). The combustion products of carbon, nitrogen, and hydrogen (C02,N2,and H20,respectively) were then measured by a thermal conductivity detector (TCD). The TCD response was recorded on a 1-mV recorder. The TCD response and the proportionality factors which were previously determined using the primary standard allowed the weight percent of carbon, nitrogen, and hydrogen to be calculated. The ladle was carefully removed and the platinum boat was reweighed. Any solid remaining in the boat was considered to be ash and thus the weight percent of ash was obtained. The difference in weight of the original sample from the sum of the weights of sulfur, carbon, nitrogen, hydrogen, and ash was taken to be oxygen.
Analysis Accuracy and Precision Errors made in quantitative analysis are divided into two categories, determinate (i.e., correctable) and indeterminate (i.e., limits in analytical techniques). When determinate errors have been eliminated, precision becomes the limiting factor in the accuracy of the analysis. Then, the mean for several measurements will approach the actual value. Indeterminate (random) errors may be modeled by a n o d Gaussian distribution where the 95% confidence limits are represented by the mean (actual value) plus or minus 1.96 times the standard deviation (std dev). Hence, the standard deviation was used to report the accuracy of one single measurement in terms of relative percent error as f1.96 (std dev) relative error, % = x 100% (1) standardized value The standardized value is the most accurate approximation of the actual value available, such as a gas or liquid standard. Approximately 30 analyses were made with each method to indicate accuracy of the analytical methods. Table I presents the results from this statistical analysis for each sample of the measurement techniques used. Relative errors for gas analyses varied from 2 to 8%, while those for liquid and solid analysis varied from 8 to 10% and from 5 to 11%, respectively, except for ash measurements, where errors were higher. Combustion Tests Test Program. Three pulverized coal combustion tests were performed to demonstrate and evaluate the analytical methods described above. These combustor tests were performed with two types of coal. A subbituminous coal from the Belle Ayre coal mine in Gillette, WY, was used in tests 1and 2. A lignite coal from the Knife River coal mine in Beulah, ND, was used in test 3. These coals were chosen because they have been tested by other investigators (e.g., Asay, 1982) and because they differ in coal rank.
Table I. Accuracies of Analytical Methods re1 error, measd species H,S, COS!
so2
cs2 NZ
0, CH,, CO COZ Ar
3P . NH,, S 2 - , CN -
so,*C N H ash
0 S a
method
concnrange
Gas Phase GC, FPD 5-100 ppm GC, FPD GC, TCD GC, TCD GC, TCD GC, TCD GC, USD GC, USD chemiluminescent
%
6
2.5-50 ppm 70-80 mol % 5-15 mol % 1-5 mol % 5-15 mol % 0.5-8 mol % 0.5-2 mol % 100-1000 ppm
Liquid Phase ion specific 1-20 mg/L electrode titrimetry 1-100 mg/L Solid Phase Perkin-Elmer 0-80 wt % 240 Analyzer Perkin-Elmer 0-10 wt % 240 Analyzer Perkin-Elmer 0-6 wt % 240 Analyzer Perkin-Elmer 0-80 wt % 240 Analyzer Perkin-Elmer 0-5 wt % 240 Analyzer LECO sulfur 0-4 wt % analyzer
8 3 5 3 3 5 2 7
8
10 5
10 11
25 20 8
See Rees et al. (1980).
Table 11. Coal Elemental Compositions
component, wt % C H N S 0 (by difference) moistureu ash
Gillette, WY subKnife R.iver, butiminous ND lignite 65.26 4.47 1.00 0.44 9.58 14.60 4.65
53.27 3.19 1.27 1.02 9.95 25.30 6.00
ASTM D 3173 (1976) immediately prior to testing. ASTM D 3174 (1976).
a
The compositions of the test coals were determined from samples from the coal feeder just prior to each run (see Table 11) with methods previously outlined for solid phase analysis. The operating parameters for each combustor test are presented in Table 111. The coal feed rates were measured with a platform scale. The analysis of the combustion products and the use of both gas and particle tracers permitted evaluation of feed rate accuracy. Test Results. Samples were obtained from near the reactor exit (115 cm from entrance) at various radial locations between the reactor centerline and wall. The gas, liquid, and solid samples were analyzed by the techniques described above. Table IV summarizes the results obtained for the permanent gases (Ar,H2,CHI, CO, C02, 02, and N2), the sulfur and nitrogen pollutants (SO,, H2S, COS, CS2,NH,, HCN, and NO), and the char composition. CS2, which is not soluble in water, was assumed to remain in the gas phase. Part of the COS reacts with water to form HzS, a portion of which dissolves in the quench water as S2-. However, since H2S forms in much higher quantities than COS (Levy et al., 1970; Gangwal and Jayanty, 1979), any S2- in the quench water was reported
Ind. Eng. Chem. Fundam., Vol. 22, No. 3, 1983 297
content of the coal and the coal feed rate. The coal feed rate was determined indirectly from a carbon balance (accounting for carbon in CO, COP,CHI, and char), since the coal feed rate determined from the weight scale was not thought to be reliable. Sulfur quantities in the outlet gas phase were determined from the product of the measured species concentration in the gas and the molar gas flow rate. The gas flow rate was reliably determined indirectly from an argon balance, assuming steady state flow of inert argon. Average sulfur species concentrations near the combustor exit were determined by integrating local species concentrations at various radial portions over the cross-sectional area of the combustor assuming the gaseous specific mass flow rate to be uniform. The amount of sulfur in the char was determined from an elemental analysis on char obtained from the scrubber. The outlet char flow rate was determined indirectly from the coal feed rate through the use of ash as an inert trace element in the char. While ash is not entirely inert (Padia, 1976),the error thus introduced is less when high burnout levels are approached. From these values, independently determined inlet and outlet quantities of sulfur were compared. Carbon balance results gave a coal feed rate of 12.1 kg/h while the ash balance suggested a carbon burnout of 99%. Sulfur measurements in the outlet char showed that 89% of the sulfur had been driven from the coal into the gas phase. Most importantly, the mass flow rate of the sulfur entering in the coal (123 g/h) agreed to within 1% with the sulfur exiting in the gas phase (110 g/h) and in the char (13 g/h). Thus, to within the accuracy of the analytical measurements and the limitations on the mass balance, the sulfur balance was independently satisfied. Because of the complexity of the sulfur mass balance (eq 2) and the associated number of required assumptions, the observed sulfur balance agreement is possibly more accurate than can be justified. Even so, the single result does add confidence to the sulfur measurement methods. (Table VI1 contains a summary of sulfur and carbon conversion values.)
Table 111. Operating Parameters for Combustor Tests test 1 test 2
test 3
coal type subbit. subbit. lignite coal feed rate (wet), kg/h 13.6a 11.0d 12.ld stoichiometric ratioe 1.2a 1.4d 2.2d 14.6 14.6 25.3 coal moisture content, % 333 366 305 primary air temp, K C 32.2 35.1 primary air rate, kg/h 27.6 4.41 primary argon concn, mol % 4.71 5.32 589 589 589 secondary air temp, K secondary air rate, kg/h 116.4 116.4 161.2 3.2 3.2 7.0 theoretical secondary swirl no.f axial probe location, m 1.15 1.15 1.15 parameter
a From scale reading. Immediately prior to testing. At combustor entrance. d From forced carbon balance; see Harding (1980). e Stoichiometric ratio is the ratio of the available oxygen needed to effect complete combustion. f Swirl no. is a ratio relating the tangential momentum of a swirling flow and the axial momentum. Higher swirl numbers imply greater swirling flow.
as H2S. Thus, COS was considered to be strictly a gas phase species. SO, and H2S are both highly soluble. Distributions of SO, and H2S between the gas and liquid phases for testa 1 , 2 , and 3 are shown in Table V. Nearly 100% of the SO, and more than 60% of the H2S were dissolved in the quench water. Information concerning the fate of sulfur in coal was also obtained from these tests, as shown in Table VI. As expected from equilibrium calculations, SO, accounted for upwards of 90% of the gas-phase sulfur, H2S accounted for 1 to 10% of the total sulfur gases, and COS and CS2 accounted for less than 1% of the total sulfur gases. Essentially all of the NH3 and HCN dissolved in the quench water (Price, 1980). Consequently, the values reported in Table IV were determined solely from analysis of the quench water. Sulfur Balance. The purpose of the sulfur mass balance was to provide an independent evaluation of the reliability of the sulfur species measurement. A detailed sulfur mass balance was performed only on test 3, since this was the only test in which sulfur concentration was measured in the outlet char. The assumption of steadystate operation of the combustor leads to the equation
Conclusions Extensive laboratory testing has established the level of accuracy or relative errors of the analytical methods outlined herein. Relative errors for gas analyses varied from 2 to 8%, while those for liquid and solid analyses varied from 8 to 10% and from 5 to 11%, respectively, except for ash measurements, where errors were higher. A sulfur mass balance from a laboratory-scale coal reactor test agreed to within 1%, providing further verification of the validity of the methods for sulfur species analysis.
Other sulfur containing species were neglected. Sulfur in the inlet coal was determined from the product of sulfur Table IV. Gas Composition from Combustor Tests test radius, no. cm
permanent gases, mol % (dry)
pollutants, ppm (dry)
H,S COS CS, NH, HCN NO 61 1 3 3 350 1804 256 20 0 40 278 255 0 16 0 243 248 0 161 90 9 0 630 1088 304 2 33 3 0 423 489 291 9 0 0 296 306 388 9 0 0 163 285 294 9 0 0 153 230 286 3c 6 0 0 39 55 442 7 0 0 45 54 407 9 0 0 36 58 396 6 0 0 57 57 416 Char analysis, wt %: C = 47.59, H = a Char analysis, wt %: C = 44.95, H = 3.93, 0 = 11.72, N = 1.29, ash = 38.10. Exhaust scrubber char analysis, wt %: C = 11.75, H 4 0.55, 0 = 2.29, N = 0.21, 3.81, 0 = 11.23, N = 1.16, ash = 36.21. S = 2.29, ash = 82.42. 1
0 8 10 0 4 6a 8 lob 0 4 6 10
Ar
H,
CH,
CO
CO,
0,
N,
SO,
1.37 1.41 1.33 1.73 1.67 1.62 1.53 1.57 1.44 1.45 1.50 1.43
0.02 0.08 0.09 0.40 0.89 0.07 0.05 0.07 0.01 0.00 0.00 0.00
0.11 0.00 0.00 0.03 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
3.37 0.27 0.27 1.55 0.65 0.25 0.19 0.23 0.00 0.00 0.00 0.00
13.52 7.62 7.35 11.72 11.12 8.64 7.86 7.86 8.62 8.49 7.40 7.15
2.99 9.47 9.89 4.79 6.42 8.31 9.15 9.17 9.07 9.10 9.63 10.45
78.62 81.15 81.08 79.78 79.25 81.67 81.22 81.08 80.85 80.96 81.47 80.97
446 173 153 445 268 207 149 117 835 722 450 262
298
Ind. Eng. Chem. Fundam., Vol. 22, No. 3, 1983
Table V. Distribution of Sulfur Gases among Sample Phases (wt %)" species
phase
test 1
test 2
test 3
SOX SOX
gas liquid
1 99 38 62
2 98 23 77
1 99 1 99
gas H2S liquid H,S a Data were obtained by using feed rates based on forced carbon balance.
Table VI. Fate of Fuel Sulfur sulfur form
test 1
test 2
test 3
char (% S)
2.2" 84.6 10.6 1.3 0.6
13.3a 78.4 7.7 0.6 0.0
10.9 87.8 1.3 0.0 0.0
sox H*S
cos CS, a
From a forced sulfur balance.
Table VII. Summary of Sulfur and Carbon Conversion C S feed rate, C accounted S accounted kg/h conv, % for, % conv, % for, %
13.6"
46.1
Test 1 96.1d
105.2
105.2d
13.6" ll.Ob
73.1 90.9
Test 2 82.3 100.Ob
69.5 86.3
69.5d 86.3d
22.4" 12.0b 8.4c
53.2 98.9 143.0
Test 3 53.8 100.Ob 143.6
47.9 89.1 127.9
54.0 100.4 144.2
From forced carbon a From scale measurement. balance. From forced ash balance. Solid phase not measured.
This work provides a foundation for the chemical analyses of local three-phase samples taken from a coal combustion reactor with a water-quenched probe. The data thus obtained from sample analysis with these methods can provide significant insight into the processes that occur in a coal combustor. Further, the data from
these local samples throughout the combustor can be used to evaluate the utility of mathematical models that are used to predict such processes.
Acknowledgment Blaine Asay, John Highsmith, and Steven Zaugg assisted in the chemical analyses, while technician, drafting, and secretarial services were provided by Michael King, Kathleen Hartman, and Ruth Ann Perkins, respectively. This research project was supported by the EPRI under contract RP-364-2, with Mr. John Dimmer as project officer, and by the Brigham Young University Research Division. Registry No. S, 7704-34-9; SOz, 7446-09-5; H2S, 7783-06-4;
COS,463-58-1; CSZ, 75-15-0; NO, 10102-43-9;Nz, 7727-37-9;Sods, 14808-79-8; S", 18496-25-8;CN-, 57-12-5; NH4+, 14798-03-9.
Literature Cited Amerlcan Soclety for Testing and Materials, "1978 Annual Book of ASTM Standards"; ASTM: Philadelphia, PA, 1978. Asay, B. W. Ph.D. Dlssertatlon, Bdgham Young Unhrersity, Provo, UT, 1982. Baumann, E. W. Anal. Chem. 1974, 46, 1345. Budesinsky, B. W. MIcrochem. J. 1975, 20, 380. Burkinshaw, J. R. M.S. Thesis, Brlgham Young Unhrersity. Provo, UT, 1981. Gangwall, S. K.; Jayanty, R. K. M. "Sampling and Analysis of Low Molecular Weight Sulfw Compounds in Process Effluents"; Flnal Report, EPA Contract No. 88-02-215, Task 18. IndustrialEnvironmental Research Laboratory, Environmental Protection Agency, Research Triangle Park, NC, 1979. Hardlng, N. S. Ph.D.Dlssertatbn, Brigham Young UnhrersHy, Rovo. UT. 1980. Harding, N. S.; Smoot, L. D.; Hedman, P. 0. A I C h E J . 1982. 26(4), 573. Levy, A.; Merryman, E. L.; Reid, W. T. Envkon. Scl. Techno/. 1970, 4, 218. Luthy, R. G.; Bruce, S. G. "Kinetlcs and Reactlons of Cyanlde and Reduced Suifw Specles to Form Thlocyanate", DOE Report No. FE-2498-32, DOE Contract No. EX-78-S-01-2496, Carnegie-Meilon University, Pittsburgh, PA, 1978. Padla, A. S. ScD. Thesis, MIT, Cambridge, MA, 1978. Price, T. S. M.S. Thesis, Brigham Young Unhrerslty, Provo, UT, 1980. Rlce, T. D.; Smoot, L. D.;Hedman, P. 0. Ind. Eng. Chem. Fundam. 1983, 22, 110. Rem, D. P.; Smoot, L. D.;Hedman, P. 0. "Nitrogen Oxlde Formation Inside a Laboratory Puhrerized Coal Combustor"; 18th Symposium (International) on Combustion, Pbburgh, PA, 1980. Skinner, F. D.; Hedman, P. 0.; Smoot, L. D. "Mlxlng and Gasificatlon of Coal in an Entrained Flow Gaslfler"; 1980 ASME Wlnter Annual Meeting, Chicago, IL, 1980. Thwgood, J. R.; Smoot, L. D.; Hedman, P. 0. Combust. Sci. Techno/. 1980, 21, 213.
Received for review November 2, 1981 Revised manuscript received January 27, 1983 Accepted February 28, 1983