Environ. Sci. Technol. 2001, 35, 21-25
Measurements of Methane Emissions from Landfills Using a Time Correlation Tracer Method Based on FTIR Absorption Spectroscopy BO GALLE* AND JERKER SAMUELSSON IVL Swedish Environmental Research Institute, P.O. Box 47086, S-402 58 Go¨teborg, Sweden BO H. SVENSSON AND GUNNAR BO ¨ RJESSON Department of Water and Environmental Studies, Linko¨ping University, S-581 83 Linko¨ping, Sweden
Methane is an important climate gas contributing significantly to global warming. A large part of the anthropogenic emissions of methane comes from landfills. Due to the biogenic origin of these emissions and the inhomogeneous characteristics of landfills and their soil cover, these emissions show large spatial variation. Thus, development of reliable and cost-effective methods for measurements of these emissions is an important task and a challenge to the scientific community. Traditionally, field chamber methods have been used but also different area integrating methods based on downwind plume measurements. These measurements have been supported by meteorological data either directly from local measurements or by controlled release of tracer gas from the landfill providing the dispersion characteristics of the plume. In this paper we describe a method, the Time Correlation Tracer method, combining controlled tracer gas release from the landfill with time-resolved concentration measurements downwind the landfill using FTIR absorption spectroscopy. The method has been tested and used on measurements at a landfill in southern Sweden over the past 1.5 years. The method has proven to be a usable method for measurements of total methane emission from landfills, and under favorable meteorological conditions we estimate an achievable accuracy of 15-30%. The real time analysis capability of the FTIR makes it possible to judge the success of the measurement already on site and to decide whether more measurements are necessary. The measurement strategy is relatively simple and straightforward, and one person can make a measurement from a medium sized landfill (1-4 ha) within a few days to a week depending on the meteorological situation.
Introduction Methane is a climate gas contributing to the greenhouse effect. The concentration in the atmosphere increases globally by 0.6-0.8% per year (1). An important source of methane * Corresponding author phone: +46 317256200; fax: +46 317256290; e-mail:
[email protected]. 10.1021/es0011008 CCC: $20.00 Published on Web 12/02/2000
2001 American Chemical Society
is bacterial processes. Old landfills with urban waste are known to produce and emit considerable amounts of methane. The IPCC (1) has estimated that more than 10%, 20-70 Tg‚year-1, of the total anthropogenic methane emissions, 385 Tg‚year-1, originate from landfills (2). Besides developing and implementing new waste treatment practices such as incineration and separation at source, it is important to find methods to reduce the methane emissions from future and existing landfills. The organic carbon available in a landfill is biogenically degraded to CH4 or CO2, about 50% of each. Since CH4 has a GWP (Global Warming Potential) that is 20 times stronger then CO2, a considerable improvement is possible if a larger fraction of the carbon was emitted as CO2. There is in principle two ways to achieve this: by pumping the gas out and burning it or by methane oxidation via bacteria in the top soil. Both these methods are currently under strong development, and in order to validate the effect of different methods there is a need to be able to quantify the total emissions from landfills. A project has been initiated aming at, via measurements at representative sites, improving the national budget of methane emissions from landfills in Sweden as well as studying the effects of improved covering of leaking landfills. As a first step in this project, different measurement techniques for methane flux measurements were tested. A traditional and straightforward way to determine methane emissions from landfills is using field chambers (3, 4). Due to the spatial variability in the emissions, a large number of measurements are needed to characterize a site; however, still the errors are considerable. Thus, different methods, integrating the total emission from the landfill by concentration measurements in the plume downwind of the site, have been tried in combination with dispersion modeling. To constrain the model, meteorological measurements as well as controlled tracer gas releases have been used. One method that has been used, to quantify fugitive emissions from the petrochemical industry, is based on a well controlled release of a tracer gas from locations simulating the expected source distribution at the site. The plume is then sampled downwind the site by means of bags or canisters along a transect perpendicular to the wind direction. The concentration of tracer CT as well as the compound under study, CM, is then analyzed. With concentrations expressed in a mixing ratio, MT and MM being molecular weights of the respective gas, and knowing the emission rate of the tracer QT, the emission rate of the measured gas QM is obtained from eq 1.
CM‚MM CT‚MT
Q M ) QT ‚
(1)
Provided that the tracer release simulates the source well, a good estimate of the emission can be obtained. The quality of the simulation is confirmed by the variation in ratio between source gas and tracer in the different samples (5, 6). Drawbacks of the method are the rather elaborate and time-consuming sampling procedure as well as the fact that data are not available in real time, although analysis can be made on site. A variation of this method is to use mobile, sensitive, direct monitoring instruments for the plume traverses. Then real time information about the plume characteristics and thus the quality of the source simulation is available (7, 8). A technique that has proven to have a good potential for flux measurements of climate gases is LongPath FTIR spectroscopy (9-16). Long-Path FTIR has been used, combined with tracer techniques, for measurements VOL. 35, NO. 1, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
21
of emissions from area and volume sources taking advantage of its line integrating capabilities. Here, the Cross Wind Integrated (CWI) concentrations are obtained directly from the measurement. If the tracer release simulates the source well, then the emission is obtained in analogy with eq 1 (17). If the source gas emitting area is too large to be simulated by the tracer, then the tracer can be used to obtain information about the vertical diffusion characteristics. This latter approach has been applied to measurements of methane emissions from open coal mines (18), in a validation experiment using an area distributed methane source (19), in urban methane emission measurements (8), and in measurement of landfill methane emissions (20). In the present paper, we describe a novel method for measurement of total emissions of methane from landfills. The method is based on release of a tracer gas at the landfill combined with time-resolved concentration measurements downwind the landfill using Long-Path FTIR absorption spectroscopy. From time-correlative analysis of the data, the total methane emission from the landfill is determined.
FIGURE 1. Schematic overview of the closed multireflection cell system.
Experimental Section In the Time Correlation Tracer method the FTIR system is put up downwind of the site to be studied. A tracer gas is released at a known rate from the site. During a period with varying meteorology, the concentrations of the gas under study and the tracer gas are monitored. The part of the concentration time series of the studied gas, that correlates with the time series of the tracer gas, can be assumed to have its origin in the area where the tracer gas is released and can be quantified using the known tracer gas flux. If the measurements are conducted far enough from the site, the site can be regarded as a point source, and the total emission is then determined. By releasing tracer gas from different areas on the site, or using different tracer gases, the emission from different parts of the site may be determined. Besides being used to quantify total emissions of methane from landfills, the method has also been successfully applied in measurements of ethylene, propylene, and ammonia from petrochemical plants (21) as well as terpenes from the forest industry (22). For the time-resolved concentration measurements, both closed and open multireflection systems as well as an open-path retroreflector system have been used. Long-Path Fourier Transform Infrared Spectroscopy (LPFTIR) is an optical technique based on infrared absorption. Infrared light is transmitted a distance through a gas, and an absorption spectrum of the gas is recorded. The spectra of different gases are analyzed using multiregression techniques e.g. CLS (Classical Least Squares). With this algorithm, prestored calibration spectra of the studied and interfering compounds are fitted to the measured spectrum using a linear least-squares fit procedure. Thus, a number of interesting gases can be analyzed (CH4, CO, CO2, N2O, H2O, NH3, hydrocarbons etc.) simultaneously and with high time resolution. By the use of long optical path lengths, 50-1000 m, sensitivities down-to-mixing ratios of a few ppb are obtained. Long path lengths are obtained by the use of different optical arrangements. Depending on the nature of these arrangements, different measurement strategies are employed. In the present application a medium resolution (1 cm-1) FTIR spectrometer (Bomem MB100), connected to a closed multireflection cell (volume 12 L, optical path 56 m), was used. For improved signal/noise, 13 spectra were coadded before evaluation, yielding a time resolution of 1 min. A schematic view of the system is shown in Figure 1. By means of a pump and 1/2" Teflon tubing, air is pumped through the cell at a flow rate of 6 L per min. The whole system is built on an optical breadboard and housed in a thermostated container for maximum mechanical stability 22
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 1, 2001
FIGURE 2. Map of Falko1 ping municipal landfill. Also shown are the locations of the N2O tracer release cylinders. and in order to avoid humidity on the hygroscopic KBr windows. The pressure and temperature in the cell are logged together with each spectrum. The system was housed in a Volkswagen van and self-supplied with electricity from a motor generator. Thus, it could easily be moved in accordance with the prevailing wind direction. Calibration spectra of CH4, N2O, and H2O are synthetically derived from the database HITRAN (23) using the Fortran program MALT (24). As tracer gas N2O was used. The reasons for using N2O instead of the commonly used SF6 are 2-fold. First, SF6 is a very potent climate gas with a GWP (Global Warming Potential over 100 years) of 24900 as compared to 120 for N2O. Second, SF6 is measured at 960 cm-1 and requires the use of an MCT (Mercury-Cadmium-Telluride) detector. Nitrous oxide is measured at 2220 cm-1 and methane at 2950 cm-1, both covered by an InSb detector. Although methane also can be measured by the MCT detector, the InSb detector is more linear, more sensitive and less noisy, and therefore preferred. Nitrous oxide was emitted from two cylinders, separated perpendicular to the wind direction to cover the landfill. The flow was adjusted to a total release rate of 5.9 kg‚h-1 and controlled by rotameters and volume integrating meters. After each measurement the cylinders were weighed as a control of the released quantity.
Results and Discussion The Time Correlation Tracer method has been applied to measurements of methane emissions from a landfill serving a small city, Falko¨ping, in southwestern Sweden. The city has had a stable population of 32000 inhabitants since the start of the actual landfill in 1960, and during the past few years the yearly deposition of municipal waste has been 18000 tons. A methane extraction system has been operating at the landfill since the beginning of the 1990s. A first set of measurements was conducted on October 21-24, 1997. During this period the average extracted gas volume was 84 m3‚h-1 with a methane content of 57%, corresponding to a methane emission of 34 kg h-1. Figure 2 shows a map over the landfill indicating the positions of the N2O gas cylinders.
FIGURE 3. The methane concentration measured 500 m downwind the Falko1 ping landfill. Also marked is the period when the tracer gas was released from the landfill.
FIGURE 4. Time series of the mixing ratios of the tracer (N2O) and CH4 measured 500 m downwind a municipal landfill.
FIGURE 5. The relation between methane and nitrous oxide concentrations in the downwind plume. Also shown is a leastsquares fit to the data having N2O mixing ratios higher than 320 ppb. In the first experiment the FTIR-van was positioned 500 m downwind of the site in the ESE direction. Figure 3 shows the methane concentration measured during 20 h. During this period the wind direction and the wind speed were varying. During the night low windspeed and possibly inversion resulted in very high concentrations. During a 3-h period, 13.40-16.40, the tracer release system was turned on. Figure 4 shows an expanded view of the measured methane and nitrous oxide concentrations during the time period when the tracer gas was turned on. The good correlation between the two time series can be seen, indicating the absence of interfering sources and good mixing between the methane plume and the tracer. As no N2O in excess of the ambient level of 315 ppb can be seen before the tracer release is turned on at 14:20, it can be concluded that the N2O emission from the landfill is negligible. In Figure 5 the concentrations of CH4 and N2O are plotted against each other. By making a linear least-squares fit to the data points corresponding to the N2O mixing ratios above the ambient N2O value (>320 ppb), the ratio of CH4 and N2O in the plume is obtained. After multiplying by the source strength of the tracer and correcting for different molecular
FIGURE 6. The correlation between methane and nitrous oxide concentrations during a measurement on October 24. The line represents a fit to data representing overlapping plumes, as shown in Figure 7.
FIGURE 7. Methane and nitrous oxide concentrations shown scaled to each other for the measurement on October 24. weights, a total methane emission of 41 kg‚h-1 was obtained from eq 1. To perform a good measurement, the distance from source to measurement system must be large enough to ensure a good mixing between the emitted gases. This distance is dependent on the geographical and meteorological conditions. To demonstrate the effect, when these conditions are not properly fulfilled, results from a measurement on October 24 are presented. The wind was blowing from the NW, and the FTIR system was relocated to a place 600 m SE of the landfill (position B in Figure 2). In Figure 6 the correlation between the concentrations of methane and nitrous oxide is shown for the period. It is seen that part of the data from this time interval do not correlate linearly. To investigate this further, an expanded view of the time series of the concentrations of methane and nitrous oxide for the actual time interval is shown in Figure 7. From this figure, it is evident that the low correlation between the time dependence of N2O and CH4 occurs on the slopes of the mixing ratio curves. The methane curve is wider than the nitrous oxide, indicating sources outside the area covered by the tracer release tubes. It is well-known that, due to the relative ease of lateral migration, slopes of landfills are known as a preferred source of methane. A longer distance between the measurement system and the landfill, or a different location of the tracer gas cylinders, would probably have solved the problem. However, the part of the data that correlates still gives an estimate of the methane emission from the region of the landfill represented by the location of the tracer tubes. This is done with the line fitted in Figure 6. The slope of this plot yields a methane emission of 34 kg‚h-1 from the studied part of the landfill. Later the same day, another measurement was performed, under different meteorological conditions. On this occasion the concentrations correlated well, and a methane emission of 44 kg‚h-1 was obtained. Figure 8 presents the results of 1 year of measurements from the same landfill. Measurements were carried out over VOL. 35, NO. 1, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
23
FIGURE 8. The emission of methane from the landfill in Falko1 ping during 1 year, along with gas recovery rates (missing for Feb 99 due to malfunction of the monitoring equipment). 1-3 day periods every second month. The average emission over the year was found to be 38 kg‚h-1. From the tracer weighted standard deviations of the individual measurements a precision of (15% was estimated, obtained according to the procedures in Mellqvist et al. (21). In this paper, simulation of diffuse leakages inside petrochemical plants indicated typical accuracies ranging between 15 and 30% when the Time Correlation Tracer method was used to determine fugitive leaks. The measurement situation is somewhat simpler in the landfill measurements, and therefore we estimate the accuracies to be the same or even better in the work presented here. No significant variation in emission was seen over the year. The measured emission was compared to model results. The model used was the Bingemer-Crutzen method (26), which was applied by the Swedish Environmental Protection Agency to calculate national methane budgets for Sweden in 1993 (27) and 1997 (28). As mentioned before, the annual amount of waste being landfilled at the Falko¨ping landfill has been fairly constant over the years, making it straightforward to assume a steady state yearly methane production rate. In 1997, 17360 tons of waste was landfilled in Falko¨ping. The fractions contributing to methane production, consisted of 7910 tons household waste, 230 tons park and garden waste, 85 tons slaughterhouse waste, and 400 tons of sludge. Using the same assumptions as the Swedish Environmental Protection Agency (27, 28) regarding the methane production potential in these different fractions, along with their assumption that 20% of dissolved organic carbon will not be anaerobically degraded, the methane production in the Falko¨ping landfill could be estimated at around 465 tons during 1997. This corresponds to an average methane production of 53 kg h-1. To obtain production rates from the measured emissions (38 kg h-1) and gas recovery data (9 kg h-1), the emissions must be compensated for the effect of methane oxidation in the top soil layer. Assuming an average methane oxidation of 10% (27, 28), an average methane production of 51 kg h-1 was derived, which corresponds well, within the uncertainty limits, with the model. Local meteorology may have a large impact on the daily emissions. Changes in atmospheric pressure are known to affect the emissions as well as changes of the diffusivity in the top soil layer. This was clearly observed at measurements in Falko¨ping in December 1998, where a drop in atmospheric pressure along with a simultaneous thawing of the ground resulted in an increase of the emission from 21 to 51 kg h-1 within 3 days. During this period the gas recovery rate was fairly constant, which otherwise also might induce a variation in the emission. The gas recovery rate has been fairly constant over the year, but for the event in December 98, where some tests were conducted, running the gas recovery equipment 24
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 1, 2001
at a high rate. To obtain a reliable estimation of the yearly emission and to observe seasonal changes, it is thus clear that more data than a few days every second month is needed. Measurements using a more traditional field chamber method were also conducted at the same site on three occasions during 1997, May 6 (81 points), July 2 (101 points), and October 21 (83 points) (25). These measurements were performed using 7.4 L static chambers placed in a regular 20 m grid. Samples were taken 4 times during 2-3 min and analyzed on a GC. These measurements gave consistently considerably lower emissions than the TCT method. However, these measurements showed large variation within each day as well as between the different occasions. As for example the flux measurements on October 21 (coincident with the TCT measurements) showed an individual variation between -15.2‚10-3 to 15.9 g CH4 m-2 h-1, and three stations made up 46% of the emissions. Depending on statistical approach the total emission was found to be between 2 and 9.7 kg CH4 h-1 as compared to 41 kg CH4 h-1 with the TCT method. Due to the large variation in the emissions it was concluded that the gridsize was inadequate to quantify the emissions properly, and a meaningful comparison between the two methods could not be made.
Acknowledgments We would like to thank the staff at the Falko¨ping landfill for their constant support and enthusiasm. The project was funded by the Swedish National Energy Administration. The instrument development have been supported by NUTEK.
Literature Cited (1) Houghton, J.; Meira Filho, L.; Callander, B.; Harris, N.; Kattenberg, A.; Maskell, K. Climate Change 1995. The Science of Climate Change. Contribution of WGI to the Second Assessment Report of the Intergovernmental Panel on Climate Change; Cambridge University Press: Cambridge, UK, 1996. (2) Prather, M.; Derwent, R.; Ehhalt, D.; Fraser, P.; Sanhueza, E.; Zhou, X. Radiative Forcing of Climate Change; Climate Change 1994. Radiative Forcing of Climate Change and the Evaluation of the IPCC IS92 Emission Scenarios; Houghton, J., et al., Eds.; Cambridge University Press: Cambridge, UK, 1995. (3) Whalen, S.; Reeburgh, W. Global Biochemical Cycles 1988, 2, 399-409. (4) Bo¨rjesson, G.; Svensson, B. Waste Manage. Res. 1997, 15, 3354. (5) Sivertsen, B. J. Air. Pollut. Cont. Assoc. 1983, 33(4). (6) Howard, T.; Lamb, B.; Bamesberger, W.; Zimmerman, P. J. Air Waste Manage. Assoc. 1992, 42(10), 1336-1344. (7) Czepiel, P. et al. J. Geophys. Res.-Atmos. 1996, 101(D11), 1671116719. (8) Lamb, B.; McManus, J.; Shorter, J.; Kolb, C.; Mosher, B.; Harriss, R.; Allwine, E.; Blaha, D.; Howard, T.; Guenther, A.; Lott, R.; Siverson, R.; Westberg, H.; Zimmerman, P. Environ. Sci. Technol. 1995, 29(6), 1468-1479. (9) Christensen, S.; Ambus, P.; Arah, J. R. M.; Clayton, H.; Galle, B.; Griffith, D. W. T.; Hargreaves, K. J.; Klemedtsson, L.; Lind, A.M.; Maag, M.; Scott, A.; Skiba, U.; Smith, K. A.; Welling, M.; Wienhold, F. G. Atmos. Environ. 1996, 30(24), 4183-4190. (10) Galle, B.; Klemedtsson, L.; Griffith, D. W. T. J. Geophys. Res.Atmos. 1994, 99(D8), 16575-16583. (11) Griffith, D. W. T.; Galle, B. Atmos. Environ. 2000, 34, 10871098. (12) Prieme´, A.; Christensen, S.; Galle, B.; Klemedtsson, L.; Griffith, D. W. T. Atmos. Environ. 1996, 30(9), 1375-1379. (13) Smith, K. A.; Clayton, H.; Arah, J. R. M.; Christensen, S.; Ambus, P.; Fowler, D.; Hargreaves, K. J.; Skiba, U.; Harris, G. W.; Wienhold, F. G.; Klemedtsson, L.; Galle, B. J. Geophys. Res.Atmos. 1994, 99(D8), 16541-16548. (14) Schafer, K. et al. Ber. Bunsen-Gesellschaft-Phys. Chem. Chem. Physics 1995, 99(3), 405-411. (15) Griffith, D.; Jamie, I.; Beasley, P.; Denmead, O.; Leuning, R.; Galbally, I.; Meyer, C. International Conference on Fourier Transform Spectroscopy; 1997. (16) Galle, B.; Klemedtsson, L.; Bergqvist, B.; Nyqvist, K.; Griffith, D. W. T. Atmos. Environ. 2000, 34, 4907-4915.
(17) McLaren, S. Ph.D. Dissertation, University of Denver, Colorado, 1995. (18) Kirchgessner, D. et al. Chemosphere 1993, 26(1-4), 23-44. (19) Piccot, S.; Masemore, S.; Ringler, E.; Srinivasan, S.; Kirchgessner, D.; Herget, W. J. Air Waste Manage. Assoc. 1994, 44(3), 271-279. (20) Savanne, D.; Arnaud, A.; Berne, P.; Burkhalter, R.; Cellier, P.; Gonze, M.; Laville, P.; Levy, F.; Milward, R.; Pokryszka, Z.; Sabroux, J.; Tausiede, C.; Tregoures, A. Third Swedish Landfill Research Symposia; 1998. (21) Mellqvist, J. Ph.D. Dissertation, Chalmers University of Technology, Gothenburg, Sweden, 1999. (22) Galle, B. Ph.D. Dissertation, Chalmers University of Technology, Gothenburg, Sweden, 1999. (23) Rothman, L. S. e. al. J. Quant. Spectrosc. Radiat. Transfer 1998, 60(5), 665-710.
(24) Griffith, D. Appl. Spectrosc. 1996, 50(1), 59-70. (25) Bo¨rjesson, G.; Danielsson, Å.; Svensson, B. H. Environ. Sci. Technol. 2000, 34(18), 4044-4050. (26) Bingemer, H. G.; Crutzen, P. J. J. Geophys. Res. 1987, 92(D2), 2181-2187. (27) Naturvårdsverket. SNV Report 4271; Swedish Environmental Protection Agency: Solna, Sweden, 1993 (in Swedish). (28) Montelius, M. SNV pm 1997.02.05; Swedish Environmental Protection Agency: Solna, Sweden, 1997 (in Swedish).
Received for review March 15, 2000. Revised manuscript received September 18, 2000. Accepted October 12, 2000. ES0011008
VOL. 35, NO. 1, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
25
