Organic Emissions from Combustion of Pine ... - ACS Publications

Mar 20, 1996 - ... times between 0.5 and 3.0 s, exit temperatures between 300 and 950 °C, equivalence ratios ((fuel/air)act/(fuel/air)stoic) between ...
0 downloads 0 Views 162KB Size
Energy & Fuels 1996, 10, 299-304

299

Organic Emissions from Combustion of Pine, Plywood, and Particleboard Jeffrey M. Hoerning,* Michelle A. Evans, Danny J. Aerts, and Kenneth W. Ragland Department of Mechanical Engineering, University of WisconsinsMadison, Madison, Wisconsin 53706 Received October 2, 1995. Revised Manuscript Received January 15, 1996X

Furnace out emissions of benzene, toluene, and formaldehyde during the combustion of wood were measured using a well-controlled plug flow research combustor which simulates an updraft fixed grate combustor. The woods examined were southern pine, southern pine plywood, and southern pine particleboard. The range of conditions for combustion were residence times between 0.5 and 3.0 s, exit temperatures between 300 and 950 °C, equivalence ratios ((fuel/air)act/(fuel/ air)stoic) between 0.1 and 0.3. For residence time of 1 s, benzene, toluene, and formaldehyde emissions were very low to undetected at exit temperatures above 650 °C, but at temperatures less 650 °C emissions increased to as high as 2 ppmv for benzene, 0.4 ppmv for toluene, and 40 ppmv for formaldehyde. Furnace out emissions and emission factors were correlated to CO concentration. For plywood the CO had to increase to 4000 ppmv (corrected to 7% O2) before emission levels increased rapidly, and for pine the CO had to increase to 2000 ppmv (corrected to 7% O2) before emission levels increased rapidly.

Introduction Large amounts of wood, wood waste, and manufactured wood products are burned on grates to produce process heat and electricity in industrial boilers. Industrial combustion chambers generate high temperatures for relatively long residence times; however, emissions of volatile organic substances are of concern if the residence time, temperature, and turbulence are inadequate. The typical industrial plant does not measure volatile organic emissions, but does measure oxygen and carbon monoxide concentrations. With sufficient excess O2 and low enough CO, the organic emissions are thought to be acceptably low. For example, the Wisconsin Department of Natural Resources (DNR) specifies good combustion practice as maintaining CO less than 600 ppmv corrected to 7% oxygen plus furnace exit temperatures greater than 675 °C for 1.0 s.1 The purpose of this project was to determine the adequacy of this criterion with respect to volatile organic emissions for selected wood and manufactured wood products in a laboratory combustor of known flow and temperature characteristics. Manufactured wood products contain wood, wood fiber, and non-wood additives such as adhesives, wood preservatives, and fire retarding chemicals. Secondary manufacturing processes can add plastic overlays, paints, varnishes, lacquers, fillers, strength additives, and dyes. Residue is in the form of edge trimmings, sawdust, sander dust, shavings, and fiber sludge which could be used to replace fossil fuels for steam generation at the industrial site. Eventually the manufactured products are discarded, and rather than going to a landfill, the wood products may be burned in a boiler. X Abstract published in Advance ACS Abstracts, February 15, 1996. (1) Wisconsin Administrative Code, Air Pollution Rules of the State of Wisconsin Department of Natural Resources; Chapter NR 445.

0887-0624/96/2510-0299$12.00/0

Previous studies that looked at the pyrolysis products of sweet gum hardwood and lignin by Nunn2,3 show that the pyrolysis products can contain up to 2 wt % formaldehyde. The particleboard we selected was made from sawdust and contained 6% phenol formaldehyde resin solids. The plywood contained 2.5% urea formaldehyde resin. Because of the formaldehyde resin, there is concern that the formaldehyde emissions may be higher than in pure wood. Formaldehyde is considered a hazardous air pollutant by the 1990 Clean Air Act. Under proper operating conditions a wood-fired, spreaderstoker boiler has low benzene and formaldehyde emissions; however, when the temperature and/or excess oxygen are too low, these emissions can be high.4 Test Setup The experimental setup, which simulates an updraft fixed grate combustor, is a 13 cm i.d. by 5 m long chamber made from low-density Ceraform ceramic risers which are 13 cm i.d., 30 cm o.d. by 30 cm long. The insulation tubes were force fit into a long steel tube assembly which was hung vertically from a fixed platform as shown in Figure 1. A propane burner was fitted into the bottom of the combustor for startup, and a mullite honeycomb grate was mounted 60 cm above the burner. Two underfire air opposed jets were located 20 cm below the grate and two overfire air opposed jets were located 40 cm above the grate. The fuel feeder was mounted on a (2) Nunn, T. R.; Howard, J. P.; Longwell, J. P.; Peters, W. A. Product Compositions and Kinetics in the Rapid Pyrolysis of Sweet Gum Hardwood. Ind. Eng. Process Des. Dev. 1985, 24, 836-844. (3) Nunn, T. R.; Howard, J. P.; Longwell, J. P.; Peters, W. A. Product Compositions and Kinetics in the Rapid Pyrolysis of Milled Wood Lignin. Ind. Eng. Process Des. Dev. 1985, 24, 844-852. (4) Hubbard, A. J. Encouraging Good Combustion Technology at Wood-Fired Facilities in Wisconsin. Presented at the National Biofuels Conference, Newton, MA, 1992. (5) Evans, M. A. M.S. Thesis, Department of Mechanical Engineering, University of WisconsinsMadison, Madison, WI, 1994. (6) Fritz, R. A.; Hubbard, A. J. Control of Hazardous Air Emissions from Wood-Fired Boilers. Bioenergy 94 1994, 335-341.

© 1996 American Chemical Society

300

Energy & Fuels, Vol. 10, No. 2, 1996

Hoerning et al. Table 1. Type of Wood Tested and Corresponding Properties wood type southern pine southern pine plywood southern pine particleboard

cube size (mm)

sp gr

% moisture

formaldehyde resin

11 11

0.55 0.65

9 7

none urea, 2.5 wt %

16

0.8

9

phenol, 6 wt %

Table 2. Operating Conditions for the Combustor temperature of gas at sampling point, °C fuel flow, g/min residence time (based on sampling point), s underfire air flow, m3(STP)/min equivalence ratio ((fuel/air)act/ (fuel/air)stoic) longitudinal Peclet number7 Reynolds number in the combustor, based on inside tube diameter

275-900 18-60 0.5-3.0 0.3-1.0 0.1-0.3 14-56 2400-7000

CO and CO2 measurements were made with Horiba Model PIR-2000 infrared gas analyzer, one set up for CO and the other for CO2. The properties of the different types of wood that were used are given in Table 1. The range of operating conditions for the combustor are given in Table 2. Temperatures were primarily controlled by the fuel feed rate. The residence times were controlled by the underfire air flow rate. The equivalence ratios were low to keep the temperatures low while still maintaining a Reynolds number in the combustor of 24007000 for good mixing. Longitudinal Peclet numbers based on reactor length ranged from 14 to 56 which indicated intermediate amounts of backmixing.7 Figure 1. Fixed bed combustor test setup. stainless steel tee at the top of the combustor. The combustor exhaust flowed to a wet scrubber and induced draft fan. The fan maintained the system at about a negative 3 mm of water and exhausted the products to the atmosphere outside the building. Single 11 mm pine and pine plywood cubes were fed at a constant rate that ranged from 28 to 100 cubes/min, into the top of the combustor and fell quickly by gravity onto the grate. The feeding of the wood cubes is started 10-15 min before the gas sample is taken and continued feeding at a constant rate throughout the experiment (usually 20 min). The cube feed rate must be very constant to maintain constant combustion conditions. The cube feeder consisted of a vibrating hopper feeding the cubes onto a Teflon-coated stainless steel chute. Cubes were held at the bottom of the chute until a rod driven by a pneumatic cylinder pushed the cube into the combustor. A solenoid turned the air to the cylinder on and off. A pulse generator controlled the timing of solenoid, and a universal counter, which gave the time between pulses, indicated the feed rate of the cubes. Combustor gas temperatures were measured at the centerline with K type thermocouples at positions T2, T3, T4, and T5 corresponding to 0.9, 1.9, 2.5, and 3.8 m above the grate (see Figure 1). The gas temperatures were not corrected for radiation from the walls which would be cooler than the gas temperature. Typical corrections would range from 35 °C increase of actual gas temperature at the lower temperatures ranges to 50 °C at the higher temperature range, assuming worst case blackbody radiation to the thermocouple bead. The vertical temperature drop through the combustor was 50-100 °C/m depending on the conditions. Total air flowing through the combustor was measured with Teledyne Hastings-Raydist mass flow meter Model NAHL-25. The O2 measurements were made with a Beckman Industrial Model 7010 oxygen meter.

Measurement Methods For detection of volatile organic compounds during combustion, a known volume of sample gas was drawn through a desorption tube which was connected to the side of the combustor through a 200 mm long, 3 mm i.d. stainless steel transfer line. To obtain an accurate sample volume of gas, a graduated sealed 3 L acrylic tube was filled to a known level with water. During the sampling process the water was pumped out of the tube at a known rate, typically 0.1-0.2 L/min. Displacing the water in the acrylic tube was the sample gas that was drawn through the desorption tube. The difference in volume from the initial and final level of water in the acrylic tube indicated the exact volume of gas that passed through the desorption tube, typically between 0.2 and 2 L. Three types of desorption tubes were used. The first was a stainless steel glass-lined thermal desorption tube packed with 250 mg of Tenax-GR adsorbent, used for benzene and toluene detection along with PAH compounds. The second and third were solvent desorption tubes. The second tube contained activated carbon and was used to detect benzene and toluene. The third tube contained XAD-2 coated with 2-hydroxymethylpiperidine (2-HMP) and was used for formaldehyde detection (Table 3). The thermal desorption system was attached to the injection port of a Hewlett Packard 5890 Series II GC using both a Hewlett Packard 5971 mass selective detector and a flame ionization detector. The desorption heater was maintained at 300 °C for the 10 min desorption process. The sample was then cryofocused at the injector end of the GC column. The peaks were identified with the Wiley NBS library using Hewlett Packard Chemstation software. A short 0.5 m by 0.53 mm i.d. deactivated fused silica guard column was used on (7) Levenspiel, O. Longitudinal Mixing of Fluids Flowing in Circular Pipes. Ind. Eng. Chem. 1958, 50, 343-346.

Organic Emissions from Combustion of Wood

Energy & Fuels, Vol. 10, No. 2, 1996 301

Table 3. Compounds Detected with Corresponding Desorption Tube, Column Type, and Standard Method compd detected benzene and toluene

desorption tube type activated carbon (solvent desorption)

total volume sample (L)

column type HP-5 (5% phenyl)methylpolysiloxane 0.53 mm i.d., 0.88 µm film thickness, 30 m long capillary column same as above

250 mg of Tenax-GR (thermal desorption) formaldehyde XAD-2 coated with DB-Wax (polyethylene glycol) 2-hydroxymethylpiperidine 0.32 mm i.d., 0.50 µm film thickness, (2-HMP) (solvent desorption) 30 m long capillary column

detection limit

standard method

2.0

benzene 0.3 ppmv, NIOSH toluene 0.07 ppmv Method 1501

0.2

same as above

none

2.0

0.3 ppmv

NIOSH Method 2541

the injection port end of the capillary column to provide an area for cryofocusing and to protect the capillary column. For solvent desorption, NIOSH methods were followeds Method 1501 was used for benzene and toluene detection. After a known amount of sample gas was drawn through the desorption tube, the contents of the tube were added to 1 mL of carbon disulfide. A 1 µL sample was taken from this solution and injected into the GC. Method 2541 was used for formaldehyde detection. To make sure breakthrough was not occurring with the Tenax desorption tube, a second tube was added in series behind the primary tube for two test samples. Conditions were produced in the combustor that gave rise to benzene and toluene concentration that were 2-4 times higher than what is observed under normal test conditions and no breakthrough was observed. The solvent desorption tubes have a front and back section that were analyzed separately. If the back section contained more than 10% of the compound that was collected on the front section, breakthrough might have occurred and the sample was not counted. Calibration was performed every day for aromatic compounds being analyzed using certified liquid calibration test mixes. For the thermal desorption, a known amount of calibration mix was added to the desorption tube through a special fitting (from Scientific Instruments) while passing He gas through the tube for 30 min. The tube was then analyzed in the typical manner. For solvent desorption a known amount of test mix was injected into the desorption tube and then the tube was capped and let stand overnight. The tube was then analyzed in the usual solvent desorption way as described in the NIOSH methods. Data points were obtained using a 20 min sampling time which ensured that any fluctuations in emissions that might occur over that time interval would be captured. Temperatures, CO, CO2, and O2 were continuously monitored during each run at the same sample point in the combustor that the desorption tube sample was taken. These meters were calibrated daily with certified gas mixes. All the calculations are based on the time-averaged temperature and O2 level at the sample point during the test run.

Results and Discussion Preliminary tests were run using both underfire and overfire air, and using extra nitrogen mixed with the air to lower combustor temperatures. The best operation was obtained using only underfire air, which is the condition reported here. Residence time was varied from 0.5 to 3 s by adjusting the air flow and by changing the sampling location from 1.9 and 3.8 m above the grate. The combustor required about 10 min to initially warm up using the propane burner. Wood cubes were then fed at a constant rate into the combustor and the burner was shut off. After about 10 min a gas sample was taken. Steady state conditions are difficult to achieve in a small combustor of this type and require a very constant fuel feed rate. Figure 2 is typical of the conditions achieved. The emissions of volatile organic compounds from the wood cubes depended on the excess oxygen, the gas

Figure 2. Furnace temperatures and light gas emissions throughout a typical experiment.

temperature, and the residence time. Under fuel-rich conditions, high concentrations of benzene, naphthalene, acenaphthylene and anthracene were observed (Figure 3). With high excess oxygen in the combustion products,

302

Energy & Fuels, Vol. 10, No. 2, 1996

Figure 3. Chromatogram of rich fuel conditions, pine wood.

Figure 4. Benzene factor emission vs exit temperature at two residence times, for plywood.

Figure 5. Toluene emission factor vs exit temperature at two residence times, for plywood.

polynuclear aromatic compounds were below our detection limit and were not observed, but benzene, toluene, and formaldehyde were observed when the furnace exit temperature was below 650 °C. When excess oxygen was greater than 6%, benzene concentrations ranged from undetected to 2 ppmv, toluene concentrations ranged from undetected to 0.4 ppmv, and formaldehyde concentrations range from undetected to 40 ppmv as the temperature was decreased below 650 °C. The wood feed rate, the ppmv concentration of the particular compound of interest, and the total air flow were used to convert to an emission factor basis of (µg of compound)/(g of wood). The results for benzene and toluene are plotted versus exit temperature at the sampling point, for two groups of residence times and for excess oxygen above 6% in Figures 4 and 5. For a residence time of 0.5 to 1.5 s the benzene emission factor was less than 7 ( 5 µg/g of wood for exit temperatures

Hoerning et al.

Figure 6. CO concentration vs exit temperature at different residence times, for plywood.

Figure 7. Benzene emission factor vs CO concentration at two residence times, for plywood.

above 650 °C, whereas below 650 °C the emissions increased rapidly. When the residence time was increased to 2-3 s, the emissions did not increase until the exit temperature dropped below 450 °C. Toluene exhibited a similar behavior, but the emissions were lower by a factor of 2. The CO concentrations were below 1000 ppmv when the exit temperature was above 650 °C for residence times of 0.5-1.5 s (Figure 6) and above 450 °C for residence times of 2.0-3.0 s. The CO increased rapidly for exit temperatures below these values. A cross plot of the benzene emissions versus CO concentration (Figure 7) shows that the benzene is below 20 µg/g of wood when the CO is below 4000 ppmv. As the CO rises above 4000 ppmv, the benzene emissions increase rapidly. Toluene shows a similar behavior (Figure 8). Comparing formaldehyde emissions from pine plywood and pure pine (Figure 9), for pure pine with a residence time of 0.6 s the formaldehyde emissions start to increase rapidly when the exit temperature drops below 600 °C. For pine plywood with a residence time of 2.4 s the exit temperature has to fall below 375 °C before formaldehyde emissions increase. For either pure pine or pine plywood, at a residence time of 1.2 s the temperature had to drop below 500 °C for the emission of formaldehyde to increase rapidly. Figure 9 also shows that when either pure pine or pine plywood were combusted at a residence time of 1.2 s, there was little difference in formaldehyde emission. Both emission values increased at the same rate as the temperature dropped below 500 °C. Pyrolysis products of just wood can contain up to 2 wt % formaldehyde.2,3 This compares to the 2.5% urea formaldehyde resin in the

Organic Emissions from Combustion of Wood

Figure 8. Toluene emission factor vs CO concentration at two residence times, for plywood.

Figure 9. Formaldehyde emission factor vs exit temperature for different residence times.

Figure 10. Formaldehyde emission factor vs CO concentration at different residence times.

plywood plus what ever else might be contributed from the wood itself during pyrolysis. The similar behavior in emission between pine and plywood might be because most of the formaldehyde that is contained in the resin is reacted during the curing process and is no longer available as formaldehyde. For the emission of formaldehyde as a function of CO, there seem to be two separate curves, one group for plywood and one group for pure pine (Figure 10). For plywood, formaldehyde emissions increase rapidly when CO level goes above 4000 ppmv, which is the same limit found for the emission of benzene and toluene using plywood. For pure pine, formaldehyde emissions increase rapidly when the CO level goes above 2000 ppmv. Apparently the plywood produces more CO, but not more formaldehyde under similar combustion conditions.

Energy & Fuels, Vol. 10, No. 2, 1996 303

Figure 11. Formaldehyde concentration vs average temperature for particleboard, at 2 s residence time and two O2 levels.

A white smoke was observed in our sample collection chamber when sampling gases from the plywood, but not with pure pine wood. No attempt was made to identify the white smoke in this study. An unreported field study showed white smoke that was produced during the combustion of plywood to be rich in sodium which had been used as a drying agent in the adhesive. By way of comparison, separate tests were performed using Kitagawa gas detector tubes for the detection of formaldehyde. For pine plywood, concentrations of formaldehyde plus acetaldehyde were tested with 15% O2 in the combustion products and residence times of 0.9 s. At 460 °C 11 ppmv (500 µg/g) of formaldehyde plus acetaldehyde was detected, and at 530 °C 3 ppmv (130 µg/g) was detected at CO levels of 5000-6000 ppmv. These values agree with that found in Figure 9, which used the methods described in the Measurement Methods section. A thesis by Evans5 examined the emission of particleboard. The detection method used a gas sampling valve connected to a GC using a thermal conductivity detector. The column was a 5 ft long, 0.125 in. i.d. SS, coated with Hayesep-T (Alltech). The detection limit was 25 ppmv. Formaldehyde concentrations from particleboard are plotted versus exit temperature for fuel-rich conditions and for 3% O2 (Figure 11). For a 2 s residence time with rich combustion conditions, as the average temperature is reduced below 700 °C, the formaldehyde increased rapidly. At 3% oxygen the formaldehyde was less than 25 ppmv for an average temperatures to as low as 500 °C. For pure pine and plywood (Figure 9), the emission of formaldehyde increased as the exit temperature dropped below 350 °C with 2.4 s residence time and O2 concentration greater than 6%; this was based on exit temperature not average temperature. Hubbard4 measured benzene and formaldehyde emissions from a 20 million Btu/h wood-fired stoker boiler at a lumber and veneer plant. Benzene emissions ranged from 1 to 11 µg/g of wood and formaldehyde ranged from 1 to 6 µg/g of wood. CO levels corrected to 7% O2 were between 200 and 700 ppmv. Furnace exit temperatures and combustor residence time were not reported. Fritz and Hubbard6 measured CO emissions from six wood-fired boilers with heat capacities of 7-20 million Btu/h and found many instances of CO levels above 600 ppmv with some levels above 10 000 ppmv. They suggest an upper CO limit of 1080 ppmv in addition to an 8 h average limit of 600 ppmv for CO to give low emission of volatile organic compounds.

304

Energy & Fuels, Vol. 10, No. 2, 1996

In our laboratory combustor, when the CO level was above 2000 ppmv for pine and above 4000 ppmv for plywood, organic emissions started to increase rapidly. Below these two CO levels, organic emissions decreased with decreasing CO level but the relative change in emissions was very small compared to the large change above these levels. In general the increase in organic emissions seems to follow an exponential growth with respect to increasing CO. The main difference between industrial boilers and the test combustor is that the test combustor approximates a plug flow reactor. In industrial boilers the mixing is not as complete in the combustion zone. If the mixing characteristics of a boiler could be defined, then the data obtained in this experiment could be scaled to a full size boiler. Conclusion For lean combustion of plywood the benzene emission factor was less than 7 ( 5 µg/g of wood and the toluene emission factor was less than 3 ( 5 µg/g of wood for

Hoerning et al.

furnace exit temperatures of 650-950 °C and a residence time of 0.5-1.5 s. For exit temperatures below 650 °C the benzene and toluene emissions increased rapidly with decreasing temperatures. When the residence time was 2-3 s, the critical temperature for rapid increase in benzene and toluene emissions was 450 °C. Emissions of formaldehyde were less than 20 ( 10 µg/g of wood for lean combustion of plywood with exit temperatures above 500 °C and residence time of 1 s. For exit temperatures below 500 °C and for shorter residence times the formaldehyde emissions increased rapidly and emission factors of over 1000 µg/g of wood were obtained. There was little difference in formaldehyde emission between pure pine and plywood for lean combustion. Benzene, toluene, and formaldehyde emissions correlated with CO concentration, and for plywood emission remained low provided the CO was below 4000 ppmv. For pure pine, formaldehyde emission remained low when the CO was below 2000 ppmv. EF950194H