This year's edition of Emissions of Greenhouse Gases in the United States uses a new set of emissions coefficients to compute the carbon content of fuels burned in the United States. They were developed to replace, in part, the more general coefficients originally developed by Marland and Pippin and subsequently adopted by the Intergovernmental Panel on Climate Change (IPCC). (170) The IPCC coefficients were intended to be suitable for all countries, and to support the division of petroleum consumption into the products defined by the International Energy Agency (IEA). Developing U.S. national emissions coefficients makes it possible to increase the precision of U.S. carbon emissions estimates for several reasons:
This appendix describes the derivation of the emissions coefficients used in this year's report. While the discussion generally is aimed at the United States, consideration is also given to how this information might be applied to other emissions inventories.
The previous edition of EIA's greenhouse gas inventory used coal carbon emissions coefficients by coal rank based on the work of Science Applications International Corporation (SAIC). (172) Using the EIA coal analysis file, SAIC was able to identify 5,426 coal samples for which data were available on both carbon content and heat content. The sample was divided by coal rank (anthracite , bituminous , subbituminous , and lignite ) and State of production, and an emissions coefficient was calculated for each State by rank. National emissions factors by coal rank were computed as a production-weighted average of State emissions factors by coal rank.
While coal emissions coefficients vary by coal rank, coal consumption data are not actually collected by coal rank, except for the utility and manufacturing sectors. Hence, consumption by coal rank must be estimated for the industrial, commercial, and residential sectors.
For this year's report, an alternative method was adopted, based on available information: coal consumption by the State of origin and rank of coal. This computationally intensive method calculates a State coal emissions coefficient for each year, based on the "mix" of consumption by coal rank within each State. A coal emissions factor for each sector is computed on the basis of sectoral consumption by State of coal origin. This work is documented in a recent article published in EIA's Quarterly Coal Report. (173) Table A1 compares the new coal coefficients (by sector) and old coal coefficients (by rank).
The composition of natural gas , as it comes out of the ground, can be diverse. Natural gas is predominantly composed of methane (CH4), which is 75 percent carbon by weight and has an emissions coefficient of 14.2 million metric tons per quadrillion Btu. Natural gas can also contain many other compounds, which can be divided into two general categories: natural gas liquids (NGLs) and nonhydrocarbon gases.
Natural gas liquids are light hydrocarbons with boiling temperatures close to room temperature. They are typically found in vapor form in natural gas reservoirs. The most common NGLs are ethane (C2H6), propane (C3H8), and butane (C4H10). NGLs also include smaller amounts of heavier hydrocarbons, such as pentane, which are commonly described as "natural gasoline" or "pentanes plus." Because NGLs have a commercial value greater than that of methane, they are typically separated from raw natural gas at gas processing plants and sold as separate products. Ethane is typically used as a petrochemical feedstock , propane and butane have diverse uses, and natural gasoline contributes to the gasoline/ naphtha "octane pool" used primarily to make motor gasoline . NGLs also have greater density and carbon content than methane.
The amount of NGLs in raw natural gas can vary considerably. In general, "associated" natural gas, produced in conjunction with crude oil, contains more NGLs than "nonassociated" gas, produced from dedicated gas wells. Associated gas tends to strip light hydrocarbons out of the crude with which it is produced.
Raw natural gas can also contain varying amounts of nonhydrocarbon gases, including carbon dioxide, nitrogen, helium and other noble gases, and hydrogen sulfide. The share of nonhydrocarbon gases is usually less than 5 percent of the total, but there are individual reservoirs where the share can be much larger. The treatment of nonhydrocarbon gases in raw gas varies. Hydrogen sulfide is always removed. Inert gases are removed if the content is high enough to reduce the energy content of the gas below pipeline specifications. Otherwise, inert gases will usually be left in the natural gas.
In the United States, the composition of pipeline-quality natural gas is defined by limits set by interstate pipeline transmission companies. In general, pipeline-quality gas must have an energy content no lower than 970 Btu per cubic foot and no higher than 1,100 Btu per cubic foot. Hydrogen sulfide content must be negligible. Typical pipeline-quality natural gas is about 95 percent methane, 3 percent NGLs, and 2 percent nonhydrocarbon gases, of which perhaps 1 percent is carbon dioxide.
However, there is still a range of gas compositions that are consistent with pipelines specifications. The minimum emissions coefficient is for pure methane, which equates to an energy content of 1,005 Btu per standard cubic foot. Gas compositions with higher or lower Btu content tend to have higher carbon emissions factors, because the "low" Btu gas has a higher content of inert gases (including carbon dioxide offset with more NGLs), while "high" Btu gas tends to have more NGLs.
The Gas Research Institute (GRI) collected some 6,743 samples of pipeline-quality natural gas from utilities and/or pipeline companies in 26 cities located in 19 States. (174) This project was undertaken to provide information on variations in gas composition for designers of alternative fueled vehicles. Dr. William Liss kindly provided this database of samples to the EIA. Table A2 illustrates the average and median composition of natural gas samples from this database, compared with the 1976 "average" composition of natural gas used by Marland and Rotty in the research underlying the emissions coefficient used by the IPCC.
As the difference between the average and the median sample would suggest, there is some variability in the composition of pipeline-quality natural gas across the sample. Using the gas composition data from the samples, we computed emissions coefficients for each sample. Figure A1 shows the relationship between the calculated emissions coefficient for natural gas and its energy content. This figure illustrates the relatively restricted range of variation in both the energy content (which varies by about 6 percent from average) and the carbon emissions coefficient of natural gas (which varies by about 5 percent).
Figure A1. Carbon Emissions for Samples of Pipeline-Quality Natural Gas in the Gas Research Institute Database
Because natural gas purchases are made on the basis of energy content, there is a limited range of energy contents in the sample. At the same time, suppliers may produce the same energy contents with a wide variety of methane, higher hydrocarbon, and nonhydrocarbon gas combinations. Thus, the plot reveals large variations in carbon content for a single Btu value. In fact, the variation in carbon content for a single Btu value may be nearly as great as the variation for the whole sample.
The plot of carbon content also reveals other interesting anomalies. The samples with the lowest emissions coefficients tend to have energy contents of about 1,000 Btu per cubic foot. They are composed of almost pure methane. Samples with a greater proportion of NGLs (e.g., ethane, propane, and butane) tend to have energy contents greater than 1,000 Btu per cubic foot, along with higher emissions coefficients. Samples with a greater proportion of inert gases tend to have lower energy content, but they usually contain carbon dioxide as one of the inert gases and, consequently, also tend to have higher emissions coefficients.
For the full sample (N=6,743), the average carbon content of a cubic foot of gas was 14.51 million metric tons carbon per quadrillion Btu. The median carbon content for the full sample was slightly lower, at 14.47 million metric tons carbon per quadrillion Btu. For all samples with more than 1,000 Btu (N=5,849) the average carbon content was 14.47 million metric tons per quadrillion Btu. The median carbon content of this subsample was also 14.47 million metric tons per quadrillion Btu. For a narrower sample, with energy contents between 1,025 and 1,035 Btu (N=2,986), the average carbon content was 14.45 million metric tons carbon per quadrillion Btu and the median carbon content was also 14.45 million metric tons per quadrillion Btu.
In order to establish a national average emissions coefficient, it was necessary to determine the "average" composition of U.S. natural gas and the extent to which the GRI sample is representative of actual natural gas "as consumed." The EIA collects data on the energy content of natural gas consumed in the United States and publishes it in the Natural Gas Annual . (175) The reported U.S. average natural gas energy content is 1,031 Btu per standard cubic foot, varying by only 0.1 percent (1,029 to 1,033 Btu per cubic foot) over the past 5 years.
The energy content of the GRI sample is 1,027 Btu per cubic foot, a difference of 0.4 percent. In order to "weight" the GRI sample, emissions coefficients were calculated for a series of subsamples, divided by energy content (Table A3).
There is some regional variation in the energy content of natural gas. The energy content of State consumption varies from 979 Btu per cubic foot in Nebraska to 1,078 Btu per cubic foot in Utah. (176) Therefore, average emissions coefficients were calculated for ranges of energy content (Table A3). An emissions coefficient from this table was assigned to each State's natural gas consumption to construct a consumption-weighted average emissions coefficient of 14.47 million metric tons of carbon per quadrillion Btu.
As shown in Figure A1, however, there is an irreducible uncertainty of about 1.5 percent in the national average emissions coefficient. We can be confident that the national average energy content of natural gas is about 1,030 Btu per cubic foot, but the composition of gas with this energy content is consistent with a range of emissions coefficients from 14.3 to 14.7 million metric tons of carbon per quadrillion Btu.
Notes for Other Inventories. For estimates based on other than national average U.S. natural gas, the range of coefficients by Btu range shown in Table A3 can be combined with information about the energy content of the natural gas actually used to create a slightly more precise estimate than the U.S. national average.
Every year, a certain amount of natural gas is flared in the United States. Since natural gas is a valuable commodity, the country is crisscrossed by pipelines, and there is a ready market for natural gas nearly everywhere, it is not immediately obvious why natural gas is flared at all in the United States. There are several possible reasons:
Information on the energy content of flare gas, as reported by States to the EIA, indicates an energy content of 1,130 Btu per standard cubic foot. (177) A set of samples from the GRI database was used to derive typical carbon and energy contents for natural gas of this quality, producing an estimated flare gas emissions content of 14.92 metric tons per standard cubic foot. The possibility exists, however, that flare gas actually has an even higher energy content than reported in the Natural Gas Annual, since rich associated gas can have energy contents as high as 1,300 to 1,400 Btu per cubic foot.
Notes for Other Inventories. Flare gas in the United States, as noted above, is probably largely "rich" associated gas. Overseas, gas may be flared under various circumstances. Rich associated gas can be flared because there is no market for it. In this instance, the emissions coefficient should be based on a gas analysis of the associated gas. On the other hand, there are many instances in which wet gas is first run through a processing plant to extract valuable liquids for which there is a market, and then the dry "tail gas" is flared. If the gas is treated prior to flaring, then the emissions factor will be much closer to the U.S. "pipeline gas" figure.
There are no available data on the composition of "national average" raw natural gas in the United States. Based on the Natural Gas Annual computations of the amount of NGLs removed from pipeline gas, it is possible to derive an estimated energy content of raw gas of about 1,150 Btu per cubic feet. (178)
In estimating emissions coefficients for petroleum products, it is useful to determine three attributes for each product:
The density determines how many tons of material are in a known volume of petroleum products. The second and third attributes determine what portion of this material is composed of carbon.
Petroleum products vary between 9.5 degrees API gravity (the heaviest number 6 fuel oil) and 247 degrees (ethane). This is a range of 60 to 150 kilograms per barrel, or plus or minus 50 percent or so. On the other hand, the variation in carbon content is much smaller: ethane is 80 percent carbon by weight, while petroleum coke is 90 to 92 percent carbon. Thus, carbon content can vary by plus or minus 5 to 7 percent, while density can vary by 50 percent.
Although it is important to define carbon content accurately, it is more important to define density accurately. Table A4 summarizes the results of this section, with estimates of the average density, carbon content, and energy content of each petroleum product in EIA statistical series.
Crude oil and petroleum products are complex substances. Gasoline and crude oil are typically mixtures of several hundred distinct compounds, most of them hydrocarbons, containing hydrogen and carbon in various proportions. When crude oil is distilled into petroleum products, it is essentially sorted into fractions by the boiling temperature of these hundreds of compounds. Boiling temperature is strongly correlated with the number of carbon atoms in each molecule. Thus, some petroleum products have low boiling temperatures and relatively simple molecules with few carbon atoms, while others have higher boiling temperatures, larger molecules, and more carbon atoms per molecule.
The higher the boiling temperature, the greater the density of the resulting product. Greater density usually implies greater carbon content as well. Petroleum products with higher carbon contents, in general, have a lower energy content per unit weight and a higher energy content per unit volume than products with a lower carbon content. Empirical research led to the establishment of a set of relationships between density, energy content per unit weight and volume, and carbon and hydrogen content. (179) Figure A2 compares emissions coefficients calculated on the basis of the derived formula with actual emissions coefficients for a range of crude oils, fuel oils, petroleum products, and pure hydrocarbons. The actual fuel samples were drawn from diverse sources (Table A5).
Figure A2. Estimated and Actual Relationships Between Petroleum Carbon Emissions Coefficients and Hydrocarbon Density
The derived relationship is an empirical regularity based on the kinds of hydrocarbons most frequently encountered. Actual petroleum fuels can vary greatly from this relationship. In the absence of more exact information, this empirical relationship is a good place to start. Actual hydrocarbons vary from the more general relationship due to nonhydrocarbon impurities and variations in molecular structure among classes of hydrocarbons.
Most fuels contain a certain share of nonhydrocarbon material. This is primarily true of crude oils and fuel oils. The most common impurity is sulfur, which is typically from 0.5 to 4 percent by weight of most crude oils, and can form an even higher percentage of heavy fuel oils. Some crude oils and fuel oils also contain appreciable quantities of oxygen and nitrogen, typically in the form of asphaltenes or various acids. The nitrogen and oxygen content of crude oils can range from nil to a few percent by weight. Lighter petroleum products have much lower levels of impurities, since the refining process tends to concentrate all of the nonhydrocarbons in the residual oil fraction. Typically, light products contain less than 0.5 percent nonhydrocarbons by weight. Thus, the carbon content of heavy fuel oils can often be several percent lower than that of lighter fuels, due entirely to the presence of nonhydrocarbons.
Hydrocarbons can be divided into five general categories, each with a distinctive relationship between density and carbon content and physical properties. Refiners tend to control the mix of hydrocarbon types in particular products in order to give petroleum products distinct properties. In light products, the "mix" of hydrocarbon types may be more important than density in determining carbon content and emissions factors. (180) The main classes of hydrocarbons are described below.
Paraffins. Paraffins are the most common constituent of crude oil, usually comprising 60 percent or more of the total weight of the barrel. Paraffins are straight-chain hydrocarbons with the general formula CnH2n+2. Typical paraffins include ethane (C2H6), propane (C3H8), butane (C4H10), and octane (C8H18). As the chemical formula suggests, the carbon content of the paraffins increases with carbon number: ethane is 80 percent carbon by weight, octane 84 percent. As the size of paraffin molecules increases, the carbon content approaches the limiting value of 85.7 percent asymptotically.
Cycloparaffins. Cycloparaffins are similar to paraffins, except that the carbon molecules form ring structures, rather than straight chains, and consequently require two fewer hydrogen molecules than paraffins. Hence, cycloparaffins always have the general formula CnH2n. All cycloparaffins are 85.7 percent carbon by weight, regardless of molecular size.
Olefins. Olefins are a reactive and unstable form of paraffin: a straight chain with the two hydrogen atoms at each end of the chain missing. They are never found in crude oil but are created in moderate quantities by the refining process. Thus, gasoline, for example, may contain 2 percent olefins. They also have the general formula CnH2n, and hence are also always 85.7 percent carbon by weight. Propylene (C3H6), a common intermediate petrochemical product (Figure A1), is an olefin.
Aromatics. Aromatics are very reactive hydrocarbons that tend to be relatively uncommon in crude oil (typically 10 percent or less). However, light aromatics increase octane number in gasoline, and consequently are deliberately created by steam reforming of naphtha. Aromatics also take the form of ring structures with some double bonds between carbon atoms that make more hydrogen atoms superfluous. The most common aromatics are benzene (C6H6), toluene (C7H8), and xylene (C8H10). The general formula for aromatics is CnH2n-6. Benzene is 92 percent carbon by weight, while xylene, with a higher carbon number, is 90.6 percent carbon by weight. Unlike the other hydrocarbon families, the carbon content of aromatics declines asymptotically toward 85.7 percent with increasing carbon number and density.
Polynuclear Aromatics. Polynuclear aromatics are large molecules with a multiple ring structure and few hydrogens, such as naphthalene (C10H2 and 94.4 percent carbon by weight) and anthracene (C14H4 and 97.7 percent carbon). They are relatively rare and appear in heavier petroleum products.
Figure A3 illustrates the share of carbon by weight for each class of hydrocarbon. Hydrocarbons in the C2-C4 range are all natural gas liquids; hydrocarbons in the C5-C10, range predominate in naphtha and gasoline; C12-C20 comprises "middle distillates," which are used to make diesel fuel, kerosene , and jet fuel . Larger molecules generally wind up as lubricants , waxes , and residual fuel oil .
Figure A3. Carbon Content of Pure Hydrocarbons as a Function of Carbon Number
If one knows nothing about the composition of a particular petroleum product, assuming that it is 85.7 percent carbon is a not unreasonable first approximation. Since denser products have higher carbon numbers, this guess would be most likely to be correct for crude oils and fuel oils. The carbon content of lighter products is determined by the shares of paraffins and aromatics in the blend.
Unlike other fuels, the exact energy content (gross heat of combustion) of petroleum products is not generally known. The EIA estimates energy consumption in Btu on the basis of a set of industry-standard conversion factors (Table A6). These conversion factors are generally accurate to within 3 to 5 percent or so. The source of the imprecision is the variation in carbon content.
The conversion factors, and hence petroleum data denominated in energy units, do not necessarily perfectly reflect the actual composition and energy content of the underlying fuels. Hence, carbon emissions factors based on the energy content of petroleum fuels have an inherent and irreducible uncertainty. In the United States, petroleum energy data are reported on a volumetric basis. Thus, a more precise approach to estimating emissions factors would be to calculate a carbon content per unit of volume, rather than per unit of energy. Adopting this approach, however, would create a separate set of problems. The resulting coefficients would be difficult to compare with the work of other analysts.
This dilemma was resolved by calculating emissions coefficients per barrel and then dividing by the standard EIA figure for the energy content of each product, even when there was evidence that the estimate of the energy content could be improved. When these coefficients are applied to EIA energy data calculated with the standard energy content, they are equivalent to calculating emissions per barrel.
Notes for Other Inventories. Persons using these emissions coefficients for other purposes should not, in general, need to concern themselves with this issue unless they possess information based on actual measurement of the chemical composition or heat of combustion of fuels. In that case, it may be possible to increase the precision of estimates by computing a "custom" emissions coefficient. Otherwise, the precision of the emissions factors presented here is as good as the precision of the energy content estimate.
NGLs (see above) are light hydrocarbons extracted from natural gas. Liquefied petroleum gases (LPGs) are the exact same products when sold to end users. Most U.S. LPG comes from natural gas plants, but substantial amounts are also distilled from crude oil. LPG is predominantly paraffins, with very small amounts of olefins. In EIA energy statistics, LPG is divided into individual compounds, and each compound is accounted for separately. Thus, it is possible to derive an exact emissions coefficient for LPG as consumed in the United States. Table A6 shows the key characteristics of each gas.
There are several possible sources of uncertainty in this estimate. It was assumed that the average characteristics of natural gasoline are similar to hexane. Other similar products, defined in the EIA schema as "pentanes plus" and "unfractionated stream" were assumed to be identical to natural gasoline. In general, natural gasoline contains mostly pentane, followed by diminishing quantities of each category of heavier paraffin, with quantities of hydrocarbons heavier than C9 or so infinitesimally small.
EIA statistics also permit reporting the production of certain olefins related to each paraffin. For example,
ethane production could include ethylene , propane production could include propylene, and butane production butylene. However, LPG production is dominated by natural gas plant output, where olefins should not appear. Olefins are normally produced by specialized process units and should not be included as part of "oil refinery output." Incidental refinery production of olefins ought to be a fraction of refinery production of propane and butane. The output of these chemicals presumably is not large enough to affect the emissions coefficients significantly.
Considerations for Other Inventories. There is a large range of variation in the density of the different components of NGLs, compared with smaller variations in carbon and energy content. This large variation underscores the importance of calculating the density and/or energy content of the underlying products correctly. In countries without natural gas production, LPG is composed almost exclusively of propane and butane, with only small amounts of ethane. In countries without petrochemical industries, available ethane (if any) is used as refinery fuel or flared, and LPG sales consist solely of propane and butane. Thus, whether converting volumetric data into weight data, or energy data into carbon data, it is important to be aware of possible variations in the underlying mixture of products.
There are often sectoral variations in the composition of LPG consumption. LPG in the industrial sector may contain very large quantities of ethane, while LPG consumed in the residential and commercial sectors is dominated by propane.
Gasoline is the most widely used petroleum product in the United States. The EIA collects consumption data (i.e., "petroleum products supplied" by wholesalers) for several types of gasoline: leaded regular, unleaded regular, and unleaded high octane. ASTM standards permit a broad range of densities for gasoline, ranging from 50 to 70 degrees API gravity, which implies a large range of possible carbon and energy contents per barrel.
Survey data from the National Institute for Petroleum and Energy Research (NIPER) bear out the existence of a range of densities in their samples, as well as systematic variations in density between summer and winter grades of gasoline, and between gasolines with different octane ratings (Table A7). The range between the highest and the lowest observed density is about 11 percent. However, the range among averages within the sample is less: about 4.3 percent.
Having established a density for motor gasoline, the next problem is to establish a carbon content. Table A7 also shows the relatively high aromatics content of gasoline. Note that the aromatics content rises with octane number. Octane number is a measure of the variability with which a gasoline vapor ignites, using as a standard of measure the variability of iso-octane (C8H18) equal to 100. The higher the octane rating, the less likely is "engine knock" to occur in internal combustion engines. Paraffinic hydrocarbons (other than octane itself) tend to have very low octane numbers, while aromatic hydrocarbons tend to have high octane numbers. In order to boost the octane rating of gasoline, refiners typically synthesize large volumes of light aromatic hydrocarbons by steam reforming, and blend them with the largely paraffinic straight run naphtha/ gasoline (C5-C12) fraction.
Thus, that the aromatics content of gasoline rises with the octane number, as shown in Table A7, is not surprising. To calculate carbon contents, calculated analyses of the carbon content of "typical" U.S. motor gasoline were taken from the work of Mark DeLuchi. (181) DeLuchi estimates carbon contents of 86.9 percent for average "summer grades" of gasoline and 85.7 percent for "winter grades." For this report, U.S. gasoline consumption was divided into product grade and season, and emissions coefficients based on the NIPER densities, the DeLuchi carbon coefficient, and EIA's standard estimated gasoline energy content of 5.253 million Btu per barrel were assigned to each category. The share of consumption of each fuel type was used to compute a national weighted average emissions coefficient of 19.41 million metric tons per quadrillion Btu.
The uncertainty associated with this coefficient is relatively small. The possible range of carbon contents consistent with the observed range of densities and aromatics contents is no more than 2 percent or so, while, as observed above, the density uncertainty is no more than 3 percent. With more gasoline samples, it should be possible to obtain correlations between the numerous attributes of gasolines recorded in the NIPER surveys and calculated emissions coefficients. This would yield both more reliable emissions coefficients and more reliable energy content per unit weight and volume for gasoline.
Both oxygenating gasoline (as has been done in some regions of the country since 1991) and reformulating gasoline (as will be undertaken in 1995) will affect the emissions coefficient. Both programs will have the practical effect of reducing the hydrocarbon content of gasoline and increasing its oxygen content, thus reducing the carbon (and energy) content per barrel. This effect is quite minor (at the national level) for oxygenated gasoline, but will be larger for reformulated gasoline. DeLuchi estimates that reformulated gasoline is only 83.3 percent carbon by weight. (182) The emissions coefficient per Btu would probably be slightly lower.
For the purposes of the inventory, it was assumed that "motor gasoline blending components" are identical in their average composition to gasoline. This assumption will be investigated further in the coming year, since blending components may be composed of methyl tertiary butyl ether and similar compounds, rather than a gasoline-like mix.
Considerations for Other Inventories. There are regional and seasonal variations in gasoline quality in the United States, as can be demonstrated in the NIPER data. Gasoline in ozone nonattainment areas may have a considerably larger oxygen content than national average gasoline, and gasoline in cold-weather parts of the country may have more aromatics. Outside the United States, gasoline may be leaded, permitting a lower aromatics content for a given octane rating, and may have idiosyncratic characteristics related to the refinery and crude stream that produces the gasoline. However, all of these variations should affect the emissions coefficient by less than 5 percent.
Aviation gas is a relatively minor contributor to greenhouse gas emissions compared to other petroleum products, representing approximately 0.1 percent of all consumption. Aviation gasoline is used in piston-powered airplane engines. The previous EIA inventory applied the motor gasoline coefficient to aviation gasoline. However, there are some important differences between aviation gasoline and motor gasoline. While the average motor gasoline has an aromatics content of 20 to 30 percent by weight, the average aviation gasoline is likely to contain only 10 to 15 percent aromatics. The ASTM standards for boiling and freezing points in aviation gasoline effectively limit the aromatics content to a maximum of 25 percent (ASTM D910). (183) Because weight is critical in the operation of an airplane, aviation gas must have as many Btu per pound (implying a lower density) as possible, given other requirements of piston engines such as high anti-knock quality.
A carbon coefficient for aviation gasoline was calculated on the basis of the EIA standard Btu content (5.048 million Btu per barrel). This implies a density of approximately 69 degrees API gravity or 5.884 pounds per gallon. (184) To estimate the share of carbon in the fuel, it was assumed that aviation gasoline is 87.5 percent iso-octane, 9.0 percent toluene, and 3.5 percent xylene. The maximum allowable sulfur content in aviation gasoline is 0.05 percent, and the maximum allowable lead content is 0.1 percent. These amounts were judged negligible and excluded for the purposes of this analysis. This yielded a carbon content of 85 percent and a carbon emissions coefficient of 18.87 million metric tons per quadrillion Btu.
The uncertainty with respect to the composition of aviation gasoline is somewhat larger than for motor gasoline, because there are no sample data that would tend to indicate the actual average density.
EIA energy statistics distinguish between two classes of jet fuel: "naphtha-based" jet fuels and "kerosene-based" jet fuels. Naphtha-based jet fuels, used almost entirely by the military, account for less than 10 percent of total consumption. (185) Kerosene-based jet fuel is believed to consist predominantly of civil-grade Jet A (used by commercial airliners) and its military version, JP5. Other kerosene-based jet fuels include the military JP7 and JP8. Naphtha-based jet fuel products include civil Jet B and the military grades JP1, JP3, and JP4. There is considerable variation in density and carbon content across the various types of jet fuel, but actual consumption is believed to consist largely of Jet A and JP5.
NIPER surveys of jet fuels indicate a density of 42.1 degrees API gravity for Jet A (Table A7). The military counterpart, JP5, has a slightly higher average density of about 41 degrees API gravity. Naphtha-based JP4 averages about 54 degrees API gravity.
Further evidence on the density and carbon content of jet fuel comes from the work of Martel and Angelo. (186) Their work was aimed at developing methods of estimating the hydrogen content of jet fuel from more commonly available information. (Hydrogen content significantly affects the formation of smoke in jet exhaust, which makes military aircraft easier to detect.) They studied several hundred samples of jet fuel of various grades, with average densities for Jet A and JP5 similar to the results of the much later NIPER survey. They estimated an average hydrogen content for Jet A at 13.6 percent, with a range of 13.0 to 14.5 percent.
To more accurately reflect the heterogeneous nature of U.S. jet fuels, separate carbon coefficients were calculated for kerosene and naphtha jet fuels. The coefficients were then weighted by national consumption figures. EIA's standard Btu content of 5.67 million Btu per barrel was adopted for kerosene jet fuels. (187) A density of 42 degrees API gravity or 6.799 pounds per gallon was adopted for kerosene jet fuels, based on NIPER aviation turbine fuel surveys. Carbon content was estimated at 86.3 percent, based on an estimated hydrogen content of 13.6 percent (Martel and Angello) and an assumed sulfur and other nonhydrocarbon content of 0.1 percent from a series of jet fuel samples. This carbon share yields an emissions coefficient of 19.71 million metric tons per quadrillion Btu.
For naphtha jet fuel, the EIA heat content is 5.355 million Btu per barrel. The ASTM standards define the density of Jet B as between 45 and 57 degrees API. (188) This analysis assumes a density of 49 degrees API or 6.536 pounds per gallon. This density estimate is more uncertain than that for kerosene jet fuels. However, because the emissions coefficient for jet fuel is a weighted average of kerosene jet fuel and naphtha jet fuel, and the bulk of consumption is kerosene jet fuel, the impact of that uncertainty is limited. Naphtha jet fuel was estimated to contain 85.8 percent carbon, based on an estimated hydrogen content of 14.1 percent (Martel and Angello) and an assumed sulfur and other impurity content of 0.1 percent. The resulting carbon coefficient is 19.95 million metric tons per quadrillion Btu.
The coefficients for kerosene jet fuel and naphtha jet fuel were weighted by total Btu supplied for each fuel, yielding a jet fuel emissions coefficient of 19.74 million metric tons per quadrillion Btu. An alternative calculation was made using Btu contents derived directly from the fuel densities. This calculation showed a nearly identical emissions coefficient of 19.73 million metric tons per quadrillion Btu.
Distillate fuel is a common and ubiquitous product, available in many different grades for particular applications. Number 1 fuel oil is generally comparable to premium high-speed diesel, while Number 2 fuel oil (commonly used as home heating oil) is also a typical transportation and off-road vehicle fuel. Number 4 fuel oil is more comparable to heavy marine diesels. In general, EIA energy statistics distinguish between grades of distillate only for sales to "stationary combustion" customers. They do not distinguish grades for sales to transportation customers, though they do distinguish between classes of customers. About 69 percent of distillate consumption is accounted for by transportation and off-highway vehicle use, with the remaining 31 percent accounted for by residential, commercial, utility, and industrial use. (189) For the stationary applications, 96 percent of consumption is accounted for by Number 2 fuel oil and Number 2 diesel, with the most of the rest accounted for by the lighter Number 1 fuel oil.
NIPER surveys of Number 1 fuel oil averaged 42.9 API gravity, while Number 2 fuel oil averaged 33.9 API gravity (Table A7). The NIPER diesel fuel survey distinguishes between "1-D" diesel, with an average density of 42.9 API gravity, and "2-D" diesel fuel, with an average density of 33.7 API gravity. The vendors of the samples indicated that the 2-D diesel was intended primarily for use by trucks.
For the purposes of calculating an emissions coefficient, it was assumed that distillate fuel could be typified by Number 2 fuel oil. The EIA standard Btu content is 5.825 million Btu per barrel, which implies, using the National Bureau of Standards formula, a density of 35.5 degrees API or 7.064 pounds per gallon. The estimated average carbon content of diesel (86.337 percent by weight) was derived from 11 ultimate analyses of various diesel fuels obtained from several sources (Table A5). The average density of the samples was within 0.5 degree of the estimated average diesel density. The resulting carbon content coefficient is 19.95 million metric tons per quadrillion Btu.
There are two key uncertainties in this analysis: the first is the relative share of different grades of distillate fuel in transportation sector fuel consumption. If Number 1 diesel accounted for a substantial portion of total diesel consumption, the coefficient would be reduced by several percent. The uncertainty range of the carbon content is about 1 percent (85 to 87 percent carbon), and would be lower if a larger sample set could be obtained.
Residual fuel is what is left over when all the light products have been distilled from crude oil. Most impurities in crude oil end up concentrated in the residual fuel, and its density can vary considerably depending on the density of the crude oil from which it is refined and the degree of processing applied. Consequently, residual oil could potentially be one of the most difficult products in the petroleum slate to characterize.
In the United States, however, this problem is reduced to some degree, because electric utilities purchase about 37 percent of the residual oil consumed in the United States. (190) A somewhat larger share is used for vessel bunkering, and the balance is used in the commercial and industrial sectors. The residual oil (defined as Number 6 fuel oil) consumed by electric utilities has an energy content of 6.287 million Btu per barrel and an average sulfur content of 1 percent. (191) Using the Bureau of Standards formula, this implies a density of about 17 degrees API.
The NIPER fuel oil survey also covers Number 6 fuel oil and shows an average density of 11.3 API gravity (Table A7). Anecdotal evidence suggests that marine residual fuel is also very dense, with typical gravity of 10.5 to 11.5 degrees API. (192) The difference between the density implied by the energy content of utility fuels and the density observed in the NIPER surveys is probably due to nonsulfur impurities, which reduce the energy content without greatly affecting the density of the product. Impurities of several percent are commonly observed in residual oil.
To derive a carbon coefficient, the EIA standard Btu content for residual fuel of 6.287 million Btu per barrel was used. The carbon content of U.S. residual oil is estimated at 85.67 percent by weight, based on the average of ultimate analyses of a dozen samples of residual oil obtained from various sources. The emissions coefficient is 21.49 million metric tons of carbon per quadrillion Btu. If the Btu content of the residual oil samples is used, rather than the EIA standard Btu content, the coefficient falls to 21.29 million metric tons of carbon per quadrillion Btu.
While it is likely that the average density of U.S. residual oil is on the order of 11 degrees API, it is by no means certain. The carbon content of residual oil is highly variable because of the large quantity of impurities that may be present. The residual uncertainty is on the order of 3 percent, with the current coefficient probably toward the high end of the range of possible coefficients.
Notes for Other Inventories. Users outside the United States are cautioned that foreign residual oils may have considerably different characteristics from U.S. residual oil. In countries with relatively unsophisticated refining systems, residual oil gravities could rise to 24 degrees API or so, implying much less carbon per barrel and somewhat lower emissions coefficients. If refineries are running heavy, high-sulfur crudes such as Mexican Maya, the resulting residual oil can have 4 to 6 percent sulfur by weight.
Asphalt is used to pave roads. It is derived from a class of hydrocarbons called "asphaltenes," abundant in some crude oils but not in others. Asphaltenes have oxygen and nitrogen atoms bound into their molecular structure, so that they tend to have lower carbon contents than other hydrocarbons. Separate statistics for road oil were kept by the Bureau of Mines prior to 1975. Consumption was negligible when the series was discontinued. The EIA standard Btu content for asphalt is 6.636 million Btu per barrel. The ASTM petroleum measurement tables show a density of 5.6 degrees API or 8.605 pounds per gallon for asphalt. Ultimate analyses were obtained for a dozen samples of airblown asphalts, which indicated a carbon content of 83.5 percent. This generates an emissions coefficient of 20.62 million metric tons per quadrillion Btu.
Lubricant consumption is dominated by motor oil for automobiles, but there is a large range of product compositions and end uses within this category. No attempted was made to establish the shares of the various lubricants consumed in the United States. The ASTM Petroleum Measurement tables give the density of lubricants at 25.6 degrees API. (193) Ultimate analysis of a single sample of motor oil gives a carbon content of 85.8 percent. The EIA Btu content of lubricants is 6.065 million Btu per barrel. These factors give an emissions coefficient of 20.24 million metric tons per quadrillion Btu.
EIA energy statistics distinguish between two different kinds of petrochemical feedstocks: those with a boiling temperature below 400oF, generally called "naphtha," and those with a boiling temperature greater than 401oF. Petrochemical feedstocks are not so much distinguished on the basis of chemical composition as on the identity of the purchaser, who may be presumed to be a chemical company or a petrochemical unit colocated on the refinery grounds. This produces a considerable degree of uncertainty about the exact composition of petrochemical feedstocks.
The emissions coefficient for petrochemical feedstocks is calculated as a weighted average of the coefficients for naphtha and other feedstocks: 19.37 million metric tons per quadrillion Btu. The derivation of the two components is described below.
Naphthas with Boiling Temperature Less Than 400oF. The composition of petrochemical naphthas is particularly uncertain because of the sophistication of the U.S. refining industry. The naphtha fraction is basically the raw material for making gasoline. Straight-run naphtha from distillation might be 65 degrees API gravity, composed largely of paraffinic hydrocarbons, and be perhaps 84 percent carbon by weight. Straight-run naphtha in the United States is usually reformed to raise its aromatics content, releasing excess hydrogen and raising the carbon content to 86 or 87 percent. Finally, naphtha is often fed to a "BTX" chemical plant to make the aromatics benzene, toluene, and xylene, with a carbon content of 90 percent or so. In principle, a BTX plant, even when owned by the same company and colocated at a refinery, should be considered a chemical plant and hence a recipient of raw feedstock rather than a part of the refinery. However, it is possible that some refiners include their BTX plants as part of the refinery and count aromatics sales as petrochemical feedstock sales.
For this year's inventory, it was assumed that the naphtha sold to the chemical industry is straight-run naphtha, while reformed naphtha is used to make gasoline. The EIA Btu content of naphtha is 5.248 million Btu per barrel. Ultimate analyses of five naphtha samples were obtained, with an average carbon content of 84.11 percent and an average density of 67.1 degrees API. This implies a carbon coefficient of 18.14 million metric tons per million Btu.
Petrochemical Feedstocks with Boiling Temperature Greater Than 401oF. If the composition of petrochemical naphthas is uncertain, the definition of this group is virtually unknown. The boiling temperature puts the content of this product into the "middle distillate" fraction. Apparently, the only market for this fraction is gas oil feedstocks for ethylene crackers. The EIA standard Btu content for this product is identical to that for distillate fuel. This product is assumed to be identical to distillate fuel, with an emissions coefficient of 19.95 million tons carbon per quadrillion Btu.
Kerosene, formerly widely used as a lamp oil, is now a minor fuel drawn from the same petroleum fraction used to make jet fuel. It was assumed that kerosene is generally comparable to Number 1 fuel oil, with an average density of 41.4 degrees API and an average carbon content of 86.1 percent. With an EIA standard energy content of 5.67 million Btu per barrel, this implies a carbon coefficient of 19.72 million metric tons per quadrillion Btu.
Petroleum coke is the solid residue of the extensive processing of crude oil. It is a coal-like solid, usually with a carbon content greater than 90 percent, that is used as a boiler fuel and industrial raw material. The ASTM ascribes a density of 9.543 pounds per gallon. Ultimate analyses of two samples of petroleum coke average 92.3 percent carbon. (194) The EIA energy content for petroleum coke is 6.024 million Btu per barrel. These factors imply a carbon emissions coefficient of 27.85 million metric tons carbon per quadrillion Btu. The EIA energy factor, however, may be understated. Calculating the carbon coefficient using an energy content computed with the Boie Method yielded a much lower (26.09) carbon factor.
Special naphtha is defined as a light petroleum product to be used for solvent applications, including commercial hexane and all products conforming to ASTM standards D1836 and D484. (ASTM standard D484 was superseded by D235 in 1983.) D1836 defines commercial hexane. D235 includes four classes of solvent: stoddard solvent, used in dry cleaning; high flash point solvent, used as an industrial paint because of its slow evaporative characteristics; odorless solvent, most often used for residential paints; and high solvency mineral spirits, used for architectural finishes. These products differ in both density and carbon percentage, requiring the development of multiple coefficients.
Hexane is the easiest of the special naphthas for which to calculate a carbon emissions coefficient. Hexane is a pure paraffin containing 6 carbon atoms and 14 hydrogen atoms. Thus, it is 83.7 percent carbon. Its density is 76.6 degrees API or 5.649 pounds per gallon. Carbon contents for the other solvents are more difficult to calculate. Most widely used commercial solvents are compounds with between 7 and 12 carbon atoms. For the purpose of this analysis, all solvents except for odorless were assumed to have 9 carbon atoms. Odorless solvent was assumed to have 10 carbon atoms in its compounds because as the number of carbon atoms increases, the solvent has a decreasing odor.
The proportion of aromatics in each solvent is provided by Boldt and Hall. (195) Stoddard and high flash point solvents contain 15 percent aromatics, high solvency mineral spirits contain 30 percent aromatics (solvency increases with aromatics content), and odorless solvents contain less than 1 percent aromatics. All nonaromatic compounds are assumed to be paraffinic. These assumptions yield carbon contents of 84.44 percent for stoddard solvent, 84.70 percent for high flash point, 84.51 percent for odorless, and 85.83 for high solvency due to its large share of aromatics. Densities for the solvents were also drawn from Boldt and Hall. Stoddard solvent is 47.9 degrees API, high flash point 47.6 degrees, odorless 55 degrees, and high solvency 43.6 degrees.
Calculated carbon coefficients for these special naphthas are 17.17 million metric tons per quadrillion Btu for Hexane, 20.11 million metric tons per quadrillion Btu for stoddard solvents, 20.17 million metric tons per quadrillion Btu for high flash point solvents, 19.41 million metric tons for odorless solvents, and 20.99 million metric tons per quadrillion Btu for high solvency mineral spirits. EIA reports only a single consumption figure for special naphthas and estimates a heat content of 5.248 million Btu per barrel for that fuel. The densities and the carbon contents of the five special naphthas are weighted according to the following formula: approximately 10 percent of all special naphtha consumed is hexane; the remaining 90 percent is assumed to be distributed evenly among the four other solvents. The resulting emissions coefficient for special naphthas is 19.86 million metric tons per quadrillion Btu.
The ASTM standards define petroleum wax as a product separated from petroleum that is solid or semi-solid at 25oC. The two classes of petroleum wax are paraffin waxes and microcrystalline waxes. They differ in the number of carbon atoms and the type of hydrocarbon compounds. Microcrystalline waxes have longer carbon chains and more variation in their chemical bonds than paraffin waxes. For the purposes of this analysis, paraffin waxes are assumed to have 25 carbon atoms and microcrystalline waxes 40 carbon atoms. Paraffin waxes are assumed to be 100 percent paraffinic compounds, while microcrystalline waxes are assumed to be 50 percent paraffinic and 50 percent cycloparaffinic. The resulting carbon share for paraffinic wax is 85.23 percent, and the carbon share for microcrystalline wax is 85.56 percent. The density of paraffinic wax is estimated at 45 degrees API or 6.684 pounds per gallon from the ASTM petroleum measurement tables. The density of microcrystalline waxes is estimated at 36.7 degrees API, based on a sample of 10 microcrystalline waxes found in the Petroleum Products Handbook. (196)
A weighted average density and carbon content was calculated for petroleum waxes, assuming that wax consumption is 80 percent paraffin wax and 20 percent microcrystalline wax. The weighted average carbon content is 85.29 percent, and the weighted average density is 6.75 pounds per gallon. The EIA standard heat content for waxes is 5.537 million Btu per barrel. These procedures yield a carbon emissions coefficient for petroleum waxes of 19.81 million metric tons per quadrillion Btu.
Still gas is composed of light hydrocarbon gases that are released as petroleum is processed in a refinery. The composition of still gas is highly variable, depending primarily on the nature of the refining process and secondarily on the composition of the product being processed. Sophisticated refineries produce still gas from many different processes. Still gas can be used as a fuel or feedstock within the refinery, sold as a petrochemical feedstock, or purified and sold as pipeline-quality natural gas. In general, still gas tends to include large amounts of free hydrogen and methane, as well as smaller amounts of heavier hydrocarbons.
This report uses an estimate of the composition of still gas from the Gas Engineers Handbook. (197) This estimate indicates that still gas has a substantial percentage of heavier hydrocarbons, such as ethane and propane. The volume percent estimates in this source were converted to weight percents, and the carbon weight of a cubic foot of gas was computed. The carbon weight was divided by the 1,388 Btu per cubic foot gross heat of combustion of the sample to produce an emissions coefficient of 17.51 million metric tons of carbon per quadrillion Btu.
Prior to 1984, the EIA collected information on "petrochemical still gas," or still gas supplied to petrochemical plants. For pre-1984 data, the same coefficient was assumed as for all other still gas.
EIA energy statistics include several odd categories of petroleum products designed to ensure that reported refinery accounts "balance" and cover any "loopholes" in the taxonomy of petroleum products. These categories include crude oil, unfinished oils, and miscellaneous. Crude oil is rarely consumed directly: 1993 consumption was only 10,000 barrels per day, or 0.06 percent of U.S. oil consumption. (198) "Miscellaneous products" account for 0.26 percent of oil consumption. Unfinished oils are a balancing item with negative consumption, accounting for about 1 percent of U.S. oil consumption. For carbon accounting purposes, it was assumed that all these products have the same carbon content as crude oil.
As noted earlier, however, crude oil is a complex product. The EIA reports on the average density and sulfur content of U.S. crude oil purchased by refineries. To develop a method of estimating carbon content based on this information, ultimate analyses of 182 crude oil samples were collected. Within the sample set, average carbon content ranged from 82 to 88 percent carbon, but almost all samples fell between 84 percent and 86 percent carbon.
The density and sulfur content of the crude oil data were regressed on the carbon content, producing the following equation:
Percent Carbon = 76.99 + (10.19 * Specific Gravity)+ (-0.76 * Sulfur Content)
This equation had an R2 of only 0.35. When carbon content was adjusted to exclude nonhydrocarbon impurities, the R2 rose to 0.65. While sulfur is the most important nonhydrocarbon impurity, nitrogen and oxygen can also be significant, but they do not seem to be correlated with either density or sulfur content. Restating these results, density accounts for about 35 percent of the variation in carbon content, impurities account for about 30 percent of the variation, and the remaining 35 percent is accounted for by other factors, including (presumably) the degree to which aromatics and polynuclear aromatics are present in the crude oil. Applying this equation to the 1992 crude oil quality data (31 degrees API and 1 percent sulfur) produces an estimated carbon content of 85.1 percent.
Applying the density and carbon content to the EIA standard energy content for crude oil of 5.8 million Btu per barrel produced an emissions coefficient of 20.29 million metric tons per quadrillion Btu. The heat contents of unfinished oils and miscellaneous, as reported by EIA, are 5.825 and 5.796 million Btu per barrel respectively. This translates to carbon content coefficients of 20.21 and 20.31 million metric tons per quadrillion Btu.
Notes for Other Inventories. Crude oil coefficients are unimportant for an inventory based on end-use consumption. However, for inventories based on the mass balance approach, as recommended by the IPCC in its 1991 report, the emissions factor for crude oil accounts for most or all of the carbon emissions from petroleum. In undertaking inventories of this nature, the most important requirement is to get the density of crude oil
right. The actual energy content of crude oil is almost never known: what generally is known is the volume of oil produced and imported. Because the commercial value of crude oil depends on its quality, it is generally possible to obtain density and sulfur content information. Average crude oil density could easily vary from 20 to 40 degrees API, equivalent to 6.5 to 7.8 pounds per gallon, or plus or minus 9 percent. On the other hand, the carbon content is likely to be in the range of 84 to 86 percent, a variation of only 1 percent or so. In resolving carbon content questions, knowing the content of sulfur and other impurities will help to reduce the possibility of the occasional large error.
