Methane and carbon monoxide emissions from asphalt pavement: Measurements and estimates of their importance to global budgets

,

; Crutzen and Gidel, 1983].However, recent work by Lowe et al. [1988] and Manning et al. [1990] indicates that the contribution to CI-In from dead sources such as fossil fuels is more probably 26% of the total contribution with a range of 23-32%.Wahlen et al. [ 1989] are in reasonable agreement with this value.Their reported value is 21 -ñ 3% for the fossil carbon contribution to atmospheric methane at the end of 1987.Sackett and Barber [1988] propose that an overlooked radiocarbon dead source of atmospheric methane might be derivatives of fossil fuel carbon such as asphalts and road tars.They reported preliminary measurements of asphalt emissions of CH 4 using a laboratory test system.Tests consisted of experiments in which quartz tubes filled with asphalt were heated by exposure to UV light while emissions of CH4 and other hydrocarbons were monitored.Their results indicate that up to 15% of total carbon in asphalt in this system may eventually be released as CI-I•.Based on calculated U.S. production of asphalt they suggest that overall CHn emissions from newly produced asphalt may be as high as 5 Tg/year (Tg = 10 •2 g).They speculate that world methane production from asphalt must be even larger and point out that other petroleum-derived products with uses such as roofing and other construction applications may also release CI-I4 because of the materials' exposure to sunlight.
We have undertaken additional measurements of surface emissions of both old and new asphalt under natural conditions.Results are reported for CI-I4 emissions from both old (greater than 2 years since paved) and new (1 week to 4 weeks since paved) asphalt pavement on outdoor surfaces of roadways and parking lots.Emissions are monitored over time, along with surface temperature and radiative flux from sunlight, to arrive at mean and maximum CH4 fluxes as a function of these parameters.Companion data for CO fluxes from measured emissions are also reported.

PROPERTIES AND USES OF ASPHALT
Because most of the readers of this paper are not likely to be familiar with the science and engineering of asphalt, it will be helpful to briefly discuss asphalt properties and usage before we report any of the experimental details or data analysis.After a search of the literature and discussions with staff at the Asphalt Institute in College Park, Maryland, we learned that journals and reports document the facts regarding world and U.S. usage and physical properties of asphalt rather incompletely.Furthermore, there is no uniform practice in aspects of asphalt mixing, paving, and repaving.Therefore we have relied on first hand knowledge from scientists or engineers who are in asphalt related businesses for many of the descriptions that follow.Their comments are noted in the text.Wherever possible, we have cited published reports.
Asphalt is derived from crude petroleum oil.Petroleum is a naturally occurring complex mixture consisting predominantly of paraffin, naphthene (i.e., cycloparaffin), and aromatic hydrocarbons in variable proportions.It also contains small amounts of organic compounds containing sulfur, nitrogen, and oxygen, and even smaller amounts of some trace metals.Many crude petroleum oils contain asphalt in amounts varying from 5 to 70%.Manufacturing takes place at relatively low temperatures using one of three processes: distillation, air-blowing conversion, or extraction with solvents.The production of asphalt should not be confused with the production of coal tars, pitches, and road tars which are produced by a coal carbonization process at high temperatures.Once produced, asphalt has a relatively high thermal stability in comparison to coal tars, pitches, and road tars and it is known that during the curing and application phases, some emissions of hydrocarbons takes place [Puzinauskas and Corbett, 1978].
Asphalt materials for paving applications include asphalt cement, cutbacks, and emulsified asphalt.In paving applications it is mixed with aggregate to form asphalt concrete.Although it is not strictly correct in a technical sense, asphalt concrete is commonly called asphalt pavement.Nonpaving uses include roofing asphalt cements, emulsions, and fluxes.World production capacity of asphalt for the year 1986 (excluding the United States) was 851,622 b/cd [Kinney, 1986] where b/cd is barrels per calendar day.This is equivalent to 5.09 x 10 •ø kg for the year 1986.Because all numerical quantities for both U.S. and world data are listed in units commonly used by the oil and petroleum or asphalt industries, we have left the numerical figures in the units first quoted but have endeavored to convert to SI units as soon as practicable.For conversion of units used here, a barrel is 42 gallons and there are 8.596 pounds in a gallon of asphalt in U.S. petroleum units.Although the Oil and Gas Journal (OGJ) report [Kinney, 1986] [Anderson, 1988].Of this, asphalt cements are 83%, cutbacks are 5%, and emulsions are 12%.(For descriptions of these three types of asphalt used in paving, see Anderson [1988]). A. Anderson (personal communication, 1988)   As described in the experimental section, a teflon sheet setup was deployed to make similar tests on asphalt pavement used for parking lot surfaces and roadways.On a parking lot surface known to be about 2.5 years old, maximum emissions were lower than in the fresh asphalt cases, as shown in Table 1.CI-I4 reached a maximum of 3.9 I. tg/m2/hr at the time of maximum temperature and sunlight.This flux was higher than the previous measurement for old asphalt but well below that for fresh asphalt, although surface temperature (72øC) and UV sunlight intensity (38 W/m:) were the highest yet measured.
Because the summer of 1989 provided much higher sunlight intensity than the fall of 1988, we made two additional experiments to test fresh asphalt using the teflon sheet method.In one experiment we tested fresh (21 day old) asphalt of the kind used in all previous experiments.This is the asphalt described in detail in an earlier section as the common type of Boulder County asphalt.In spite of higher surface temperatures (73øC) and more UV sunlight intensity (40 W/m:), maximum CH 4 emissions were only 16 gg/m2/hr this time as compared to 22 gg/m2/hr for asphalt of similar age tested last fall.A likely explanation for this difference in emissions for two nearly identically-aged asphalt sections is that their usages differ since being lain down.Our tray asphalt (26 day old) was never tamped down nor driven over, while the road surface asphalt (21 day old) was immediately put into use to serve traffic needs.For the experiments with the teflon sheets, the volume error could be somewhat greater.Again, the estimate of the initial enclosure volume of 21.5 L is good to about 5%, but the wind effect during the time of enclosure could decrease the volume by possibly 20%.This is again a visual estimate.From Table 2, the value for 21 day old asphalt under the teflon sheet becomes 15.7:L-0.8I. tg/m:/hr but in the worst case may be underestimated by about 20%.
In every test we have described we are dealing with differences between accumulated source and background concentrations.These differences can be small enough that the error of the FID measurement is as much as 3%.Background air over the pavement surface typically had concentrations of about 2.00 ppmv of CI-I• while concentrations of emissions of CH n ranged from a few tenths of a ppmv above background to about 5.00 ppmv.For the smallest flux in Table 2 ( 2 as the concentrations, and hence fluxes, increase.

ESTIMATE OF ASPHALT EMISSIONS
Estimating asphalt emissions of CH n aud CO will be difficult since several assumptions are involved in extrapolating to world averages.Using figures from the preceding sections we can try to estimate total asphalt surface area in the world.This will only be a very rough estimate but it should overestimate world asphalt surface area.It will use typical or average values for figures in the calculation where those values are known to be reasonable worldwide.Where world values for figures in the calculation are less, well known estimates will be made which allow for the greatest possible asphalt emissions to occur.We will also make other estimates of the importance of asphalt pavement emissions by comparisons to some other sources of   Recent estimates for several methane sources of all types, including estimates made by compiling data such as in Table 1 for wetland areas, show that biogenic sources may range from 100-200 Tg/yr for natural weftands and 60-170 Tgtyr for rice paddies, while fossil fuel sources such as coal mining and gas drilling, venting, and transportation losses may range from 25-45 Tgtyr and 25-50 Tg/yr, respectively [Cicerone and Oremland, 1988].In recalling the estimated value for asphalt CI-h emissions of 0.01 Tg, it is evident that methane asphalt emissions are not only small compared to several biogenic sources, but to other fossil fuel sources as well.
If similar calculations are made for the CO emissions from asphalt comparisons can be made with emissions of CO from other sources.For example, the first test with the 7 day old asphalt (see Figure 2b

CONCLUSIONS
Based on our measurements we conclude that methane emissions from asphalt pavement cannot be a significant source  [1988].Because asphalt has other uses such as roofing and construction applications which may also be exposed to sunlight, they are also testing other surfaces made from petroleum products such as synthetic rubber, plastics, and spilled oil.These authors have found that some of these emit several times more CI-I4 than asphalt surfaces.Nevertheless, it would seem that the total CH• released from all of these additional potential sources must be very small as the actual global surface area of asphalt pavement exposed to sunlight must dwarf all the other mentioned petroleum-derived sources all countries other than the United States as well as the amount exported to the U.S. Therefore we must subtract the amount exported to the United States so as not to double count it.World usage then becomes 4.89 x 10 •ø kg excluding the United States.Comparing the total 4.89 x 10 aø kg to 2.84 x 10 •ø kg, one can see that the United States accounted for about one-third of the world asphalt usage in 1986.Similar calculations for other years since 1984 lead to about the same result.Therefore we will generalize the uses of asphalt by describing the procedures used in the United States and note exceptions where they are known to us.United States figures for 1987 indicate that 85% of all asphalt consumed in the United States was for asphalt pavement repavement could be as seldom as 12 to 16 year intervals, but is probably between 3 and 8 years for most U.S. roadways.Further, for U.S. asphalt usage, about 50% of emulsions (12% of all asphalt used in paving) are used for the top layer of asphalt and 50% are used to seal between layers.Emulsions are very stable and have no volatiles released over time.Cutbacks (5% of paving asphalt), which are the most volatile type of asphalt of those used for pavement, are being used less now and are not legal in high pollution metropolitan areas.Asphalt emissions were measured using three different experimental set-ups.In each case, samples were collected by withdrawing air into 20 mL volume air tight syringes.Teflon septa were used to plug the syringe after sampling to prevent exchange of sample gas with outside air while transferring them to the laboratory.Experimental conditions were monitored during the collections including surface asphalt temperature, air temperature, and UV light intensity.The UV light intensity was measured using an Eppley radiometer (spectral response range calibrated for 290-385 nm).All samples were analyzed within 2 hours of sampling against a CI-I4 standard at 3.88 ppmv (NBS SRM 1660a) and a CO standard at 9.70 ppmv (NBS SRM 2612a).Insmxment response for each compound was linear over the range of concentrations in this study and the precision of measurement is •0.01 ppmv for both CI-I4 and CO.A model HP-5880 gas chromatograph fitted with a flame ionization detector and a 5A molecular sieve column separated CI-I4 and CO for analysis.CO was passed through a methanizer consisting of a ruthenium carbonyl catalyst at 375ø(2 prior to its detection.Our initial testswere run using a fully enclosed tray of asphalt pavement which was obtained from a road paving crew along a new county road in Boulder, Colorado.The new and still hot asphalt (heated to 121ø(2 at time of paving) was placed into a stainless steel tray of dimensions 0.61 m x 0.43 m x 0.15 m to a height of about 0.08 m.This asphalt pavement is designated as Marshall Asphalt Mix Design and is 100%, 3/4-in.aggregate with a specific gravity of 2.634 for the mix and a composition of up to a maximum of 6.1% asphalt.Immediately after returning to the laboratory at NCAR, we enclosed the entire tray in a large FEP teflon bag (2 rail thickness, about 0.05 ram) and heat sealed the bag edges to make the enclosure as airtight as possible.(Tests on the transparent teflon sheet by UV-spectrophotometer show that it transmitted 95% of all UV light.)Corrections were made for air exchange by injecting butane as a tracer into the bag at the beginning of the experiment and monitoring its decrease over time.A pert in the bag allowed us to withdraw syringe samples and also to fill the bag with background air at the start of a new experiment.A measured amount of air was used to allow bag volume calculations for each experiment.Because of the relatively large volume of air surrounding the enclosed tray and the collapsible nature of the teflon enclosure, withdrawing samples had negligible effect on subsequent samples in determining methane flux values and little exchange between outside air and the bag took place through the septurn.More experiments were run using this enclosed tray system over the next few weeks.These tests were made to monitor emissions over time as the initially hot asphalt cooled and set in the tray.We investigated the effects of bright sunshine on asphalt and also the effects of laboratory heating of asphalt to typical outdoor pavement temperatures in the absence of light.External concentrations of CH4 and CO were monitored at all times during the enclosure measurements for experiments run both outdoors and indoors.In addition, CH4 and CO blanks were established for similar teflon enclosures without asphalt inside.A different experimental setup was used to monitor emissions from road and parking lot surfaces outdoors.The procedure is described below for a relatively old section of asphalt pavement in a townhouse parking lot.This pavement was about 2.5 years old and was sampled by covering a section with a large 1.31 m by 1.64 m teflon sheet.The center was raised to a height of about 0.25 m by using a dummy 1-L jar as a centerpiece under the sheet.Two meters of teflon 9 mm OD tubing were run from the central air space under the sheet to the edge of the sheet to allow us to take syringe samples.The approximate volume of the air cavity under the sheet was 21.5 L as determined by inflating the sheet with a measured background air source prior to beginning sample collections.The sheet edges were held down by wet sand to diminish air exchange into the enclosure.The leak rate was determined by doping the enclosed airspace with SF6 and monitoring its decrease over time.RESULTS Experimental data are summarized in Table 1.The maximum flux of CH 4 and CO measured during each experiment is shown with values for UV intensity and the asphalt surface and air temperatures at the time of maximum methane emissions.In each case, average fluxes over the whole time period were much smaller.The values in parentheses are maxima for the temperature and UV intensity during the respective experiments without regard to the time of maximum methane emissions.Fluxes were calculated by first determining the leak rate constant for incremental time intervals using the data from an inert tracer gas.The methane and CO production rates determined from concentration data were then corrected to account for leakage.Finally a flux was determined from the enclosure dimensions and the concentration data.In the first experiment, values for emissions from fresh (7 day old) asphalt show that methane flux peaked at about 22 gg/m2/hr while carbon monoxide flux peaked at about 2.6 mg/m2/hr in the enclosed tray.The value for CO emissions proved to be the highest obtained for any of our experiments with asphalt.The value for CI-I4 emissions was effectively equal to the highest value obtained in a subsequent experiment using relatively fresh (26 day old) asphalt.The methane emission peaked slightly before the time of maximum surface temperature (57øC) and close to the time of maximum UV sunlight (33 W/m•).Figures 1 and 2 detail results of this test (which correspond to the first day of testing listed in Table 1).Figures la and lb are CI-I4 and CO fluxes from the asphalt in the enclosure versus time.Figures 2a and 2b are surface asphalt temperature versus time and the UV radiometer reading in watts/m • versus time (where the radiometer is calibrated as 0.594 mW/cm•/mV).In the laboratory when the same asphalt sample was heated to 58ø(2 (day 21), methane and carbon monoxide emissions were about one-third as high, about 8.9 gg/m•/hr and 0.8 mg/m2/hr, respectively.When the asphalt sample was exposed to sunlight again (day 26) emissions built up to approximately the original level seen at the 7 day mark, although the asphalt surface temperature (47øC) and UV sunlight intensity (27 W/m •) were slightly lower than for the 7 day mark.These findings indicate that emissions of CI-I4 and CO are functions of both sunlight and surface temperature.
lower than for previous tests.Some of the decrease in emissions may have been due to conditions of less sunlight on the day of the test but the data clearly showed that for equal surface temperature, the older asphalt was not as productive as the fresh asphalt.This test was repeated several months later at the same location (asphalt age now about 3 years) to monitor emissions of CH4 on a day with bright sunlight and a hotter pavement surface.This time emissions Fig. la.Methane flux versus time of day for 7-day-old asphalt in enclosure (September 29, 1988).
0.45 I. tg/m:/hr), where the accumulated concentration was calculated from 2.30+.01 -2.00-J:.01,an additional error of 3% could be added to the volume error of 5%.The uncertainty in the value found by accumulating these errors is 6%, making the value 0.45+.03I. tg/m2/hr.Errors in the FID concentration measurements become smaller in determining the fluxes in Table uses about 85% of its asphalt for paving, while Canada uses only about 76% for the year 1987.The rest of the asphalt is used for applications such as roofing and other miscellaneous uses.Assuming that 85% is the world average, then 6.18 x 10 •ø kg were used as paving asphalt in 1986.Since a typical mix proportion for asphalt pavement is 6% asphalt and 94% aggregate, total asphalt pavement by weight is 1.03 x 10 n kg.Such an aggregate mix has a density of 144 lbs/ft • (about 2310 kgtm 3) and if it is laid 0.05 m thick (very thin for new pavement and about average for repavement) then 1.03 x 10 n kg will cover about 8.92 x 109 m •.This then is the maximum amount of asphalt pavement surface area covered in 1986 this is new pavement and 50% is used for repaving existing asphalt based on information in the section on properties and uses of asphalt, then the contribution to the potential asphalt surface area calculation is maximized with respect to new versus old pavement.Then assuming that repavement takes place globally every 12 years (12 years is almost certainly too long for a global average, using this figure we must keep in mind that for most of the day asphalt emissions do not approach the maximum flux used in this calculation and we do have evidence that asphalt emissions drop off significantly as the asphalt becomes old.For the purposes of this work, it is important to compare measured methane fluxes from asphalt, a fossil fuel source, with measurements from other sources of methane that have been studied previously.Table2 listsboth average flux of methane and a reasonable estimate of the total land surface area covered for several previous studies of tropical and temperate wetlands and rice paddies.As can be readily seen, the sources listed typically have fluxes several orders of magnitude larger than the maximum flux value of 0.0003 g/m2/d (22 lag/m•/hr x 12 hr) found for asphalt.The estimated land area for these sources is also larger than the estimate derived above for asphalt pavement surface area.
) had a maximum emission of 2.62 mg/m•/hr.Integrated emissions for this day result in 10.0 mg/m•/d, assuming a 14-hour day with at least some light and no nighttime emissions.Since the greater metropolitan area of Denver is about 210 km 2, one can make a liberal estimate of city asphalt emissions by assuming that 20% of the total surface area of the metropolitan area is asphalt including both combined.The recent article byCicerone and Oreroland [1988]  puts a value of 1 Tg/yr for asphalt methane emissions in perspective.the experiment.Conversations with Jack Boyers, Bud Brakey of the Asphalt Institute in Denver, Colorado, Andy Anderson of the Asphalt Institute in College Park, Maryland, Gerald Huber of the Asphalt Institute in Lexington, Kentucky, and Carl Stuka of Brannon Sand and Gravel in Denver, Colorado, were very helpful in describing the properties and uses of asphalt.We would also like to thank Jim Greenberg for help in making the GC measurements, particularly the butane tracer measurements, and Kathy Shiidmyer, a student assistant at NCAR, who took part in various aspects of the initial experimental work.This work has been supported by the National Aeronautics and Space Administration under order W-16,184.The National Center for Atmospheric Research is sponsored by the National Science Foundation.