Environ. Sci. Technol. 2002, 36, 3147-3156
Reformulated and Alternative Fuels: Modeled Impacts on Regional Air Quality with Special Emphasis on Surface Ozone Concentration BENEDIKT SCHELL,* INGMAR J. ACKERMANN, AND HEINZ HASS Ford Forschungszentrum Aachen, Su ¨ sterfeldstrasse 200, 52072 Aachen, Germany
The comprehensive European Air Pollution and Dispersion model system was used to estimate the impacts of the usage of reformulated and alternative fuels on regional air quality with special emphasis on surface ozone concentrations. A severe western European summer smog episode in July 1994 has been used as a reference, and the model predictions have been evaluated for this episode. A forecast simulation for the year 2005 (TREND) has been performed, including the future emission development based on the current legislation and technologies available. The results of the scenario TREND are used as a baseline for the other 2005 fuel scenarios, including fuel reformulation, fuel sulfur content, and compressed natural gas (CNG) as an alternative fuel. Compared to the year 1994, significant reductions in episode peak ozone concentrations and ozone grid hours are predicted for the TREND scenario. These reductions are even more pronounced within the investigated alternative and reformulated fuel scenarios. Especially, low sulfur fuels are appropriate for an immediate improvement in air quality, because they effect the emissions of the whole fleet. Furthermore, the simulation results indicate that the introduction of CNG vehicles would also enhance air quality with respect to ozone.
Introduction Besides alternative powertrains and improved exhaust gas reduction systems, reformulated and alternative fuels provide a promising opportunity to reduce urban to regional air pollution. In principle, reformulated fuels are conventional fuels but with a modified chemical composition or modified physical properties. This can be achieved by altering the alkane/alkene/aromatics ratio or by adding oxygenated components such as alcohols or ethers. In general, existing engine and vehicle technologies as well as the fueling infrastructure can be used without significant modifications. Hence, reformulated fuels have the advantage that basically the whole vehicle fleet can consequently be involved and a rather fast impact on air quality can be expected. In contrast, alternative fuels (e.g., compressed natural gas (CNG) or liquefied petroleum gas (LPG)) are replacements for conventional fuels, and their usage may require significant technical modifications of the vehicles and the development * Corresponding author phone: +49-241-9421206; fax: +49-2419421304; e-mail:
[email protected]. 10.1021/es015817m CCC: $22.00 Published on Web 06/14/2002
2002 American Chemical Society
FIGURE 1. Model configuration of the EURAD system for the coarse domain (N0) and the Nest 1 domain (N1); grid dimensions of the different model domains: N0, 61 × 51, ∆x ) 27 km; N1, 55 × 44, ∆x ) 9 km. The N2 domain covers the German federal state Nordrhein-Westfalen.
TABLE 1. Source Categories within CORINAIR-94 (SNAP-1 Level)a source category 1 combustion in energy industries 2 nonindustrial combustion 3 industrial combustion 4 production processes 5 extraction and distribution of fuels 6 solvent use 7 road transport 8 other mobile sources 9 waste treatment and disposal 10 agriculture, forestry
source type
Germany SO2 NOx NMVOC CO
elevated 68.9 22.4
0.4
2.1
elevated elevated surface surface
12.2 13.8 2.2 0.6
6.1 11.4 1.0 0.0
2.8 0.5 6.4 4.1
17.3 9.9 8.8 0.2
surface surface surface surface surface
0.0 1.7 0.6 0.0 0.0
0.0 46.1 13.0 0.0 0.0
50.8 31.5 3.5 0.0 0.0
0.0 58.2 3.5 0.0 0.0
a Source types are classified as either near surface or elevated. The percentage of each category to the national total emissions for different emitted species is given for Germany as an example.
of a new fueling infrastructure. It will also need some time to establish a fleet share of alternative fuel vehicles, which may also depend on several nontechnical conditions. However, some alternative fuels also offer a benefit in terms of greenhouse-gas emissions (1). Therefore, possible impacts on air quality due to the usage of alternative fuels will only be achieved in a longer time frame. Increased usage of alternative and reformulated fuels will not only change the NOx, NMVOC, and particle emissions on a mass basis but also possibly alter the NMVOC emission profile (i.e., the speciation in the emitted NMVOC mixture). The composition of the NMVOC mixture and the NMVOC/ NOx ratio crucially influence the ozone production. Therefore, before reformulated and alternative fuels are introduced to the market, detailed environmental impact studies should be performed. This includes exhaust emission analysis, atmospheric chemistry laboratory studies of additionally released fuel additives and their degradation products, and comprehensive air quality impact modeling. To estimate the impact of reformulated and alternative fuels on regional air quality, the comprehensive threedimensional air quality European Air Pollution and DisperVOL. 36, NO. 14, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3147
TABLE 2. Qualitative Vehicle Emission Evaluation for Reformulated and Alternative Fuels in Relation to Standard Gasolinea
The results of the scenario calculations are compared to the reference in July 1994 and a trend scenario in 2005. The assessment of the scenarios is done relative the trend scenario to minimize the influence of model uncertainties. Special emphasis was put on surface ozone concentrations, because ozone is a major photosmog compound.
Description of the Air Quality Model System
a ()) comparable; (down-right arrow) smaller emission factors; (upright arrow) higher emission factors; (v or V) considerably higher or less; ()/up-right arrow) or ()/down-right arrow) tendency toward higher or smaller emission factors; (RFG) reformulated gasoline; (M85) methanol with 15% gasoline blend; (E85) ethanol with 15% gasoline blend; (CNG) compressed natural gas; (LPG) liquefied petroleum gas; (RFD) reformulated diesel; (DMM15) diesel with 15% dimethoxymethane blend; (DME10) diesel with 10% dimethyl ether blend.
TABLE 3. Qualitative Evaluation of Reformulated and Alternative Fuels with Reference to Influence Factors for Introduction or Significant Market Penetrationa qualitative evaluation fuels
resources
range
vehicle cost2
fuel costs2
infrastructure
safety
gasoline baseline baseline baseline baseline baseline baseline RFG ) )/) )/) ) M85 )9 )/-9 E85 )/CNG + + LPG + ) diesel ) + + ) ) RFD ) + + ) ) DMM15 + ) DME10b a +, ++ (-, - -) advantageous (worse) than baseline gasoline passenger car, )/+ ()/-) tendency toward better (worse) evaluation; (9) from biomass; (2) highly variable (e.g., through tax incentives); (b) DME is a gas at standard pressure.
sion model (EURAD) has been used. A severe summer smog episode over Central Europe in July 1994 has been chosen as a reference, and several emission scenarios have been developed for the projection year 2005 considering the possible future use of reformulated and alternative fuels.
The EURAD model (2-4) in combination with the Modal Aerosol Dynamics model for Europe (MADE) (5) provides a comprehensive tool, which is capable of predicting particulate matter in addition to gas-phase concentrations of common trace gases. EURAD consists of three major components: the mesoscale meteorological model MM5 (6), the EURAD emission model EEM (7), and the chemistry-transport model CTM2 (3, 8). To take into account the specific features of European atmospheric chemistry, the EuroRADM gas-phase chemical mechanism is used which is based on the RADM2 mechanism (9, 10). Additionally, a more detailed description of the isoprene chemistry is implemented into the gas-phase mechanism (11, 12). Furthermore, the gas-phase chemistry mechanism was extended to include biogenic monoterpenes and the secondary organic aerosol model (SORGAM) has been integrated (13, 14). A more detailed description of the model system is given in the references cited here. Model Configuration and Reference Episode. A severe summer smog episode over Central Europe in July 1994 has been selected as a reference episode. This episode was characterized by high surface ozone concentrations over Western Europe, especially in the German federal state Nordrhein-Westfalen. Therefore, the coarse model domain (N0) covers Western Europe, whereas the smaller domain (N1) focuses on Nordrhein-Westfalen using nesting techniques (Figure 1). During the simulations, the information flow between the domains is one-way (i.e., boundary conditions for Nest n are delivered by Nest (n - 1)), but there is no feedback of Nest n to Nest (n - 1). The horizontal grid resolution ∆x of the N0 and N1 domains is 27 and 9 km, respectively. The vertical direction of the model covers the height region between the earth’s surface and 100 hPa using a σ-coordinate system with nonequidistant layers. The meteorological simulations are performed with 23 vertical layers. To resolve the planetary boundary layer sufficiently, about 12 layers are in the lowest kilometer and the near surface model layer has a thickness of approximately 35 m. For the chemistry-transport simulations, the number of vertical layers has been reduced to 15 but keeping the fine vertical resolution of the meteorological data in the lower layers (up to ∼600 m). The data for the free troposphere has been averaged in a mass-consistent way. The simulated episode lasts from July 21, 0000 UTC, to July 29, 0000 UTC,
TABLE 4. List of Simulations Indicating Model Domains, Source of Emission Information, and the Emission Reductions for the Source Category Road Transport within the Model Domain N1 acronym baseline TREND LSG-a LSG-b LSG-c CNG LSD Max Impact
3148
9
emission data remarks grid stucture from EMEP (15) national totals: CORINAIR 94 source categories and national totals by ref 16; updated NMVOC split (17, 19) EPEFE reduction for a fleet of Euro-2/3 vehicles (18); NMVOC split as TREND reduction for a fleet of Euro-3 vehicles (22); NMVOC split as TREND EPEFE reduction (18) for a fleet of Euro-2/3 vehicles; NMVOC split based on EPEFE (18) CNG fraction based on ref 23; NMVOC split for CNG vehicle emissions based on refs 24-26) Prognos (16); Neumann et al. (27) combination of scenarios LSG-b, CNG, and LSD; see scenarios for details
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 14, 2002
total emission reduction “Road Transport” for model domain N1
Germany: NOx - 36%; NMVOC - 48% TREND plus reduction gasoline passenger cars: NOx - 11.2%; NMVOC - 9.2% TREND plus reduction gasoline passenger cars: NOx - 11.2%; NMVOC - 30.0% TREND plus reduction gasoline passenger cars: NOx - 11.2%; NMVOC - 9.2% TREND plus reduction passenger cars: NOx - 7.6%; NMVOC - 5.6%; busses: NOx - 14.3%; NMVOC - 14.3% TREND plus reduction diesel passenger cars: NOx - 1.0%; NMVOC - 12.0%
FIGURE 2. Contribution of gasoline, diesel, and CNG vehicles and evaporative losses for NOx and VOC road traffic emissions in the model domain N1 for the different scenarios. The total amount of road traffic emissions in the model domain N1 is indicated by the solid line with diamonds in kt/year (right axis). See Table 4 for scenario definition.
1994. An initialization period of 48 h prior to this episode is used to generate self-consistent initial conditions. All results and interpretations in the following sections refer to the period July 21-July 29, 1994. Emission data used for N0 and N1 simulations are based on the EMEP emission inventory (15) and the CORINAIR-94 emission database. Country-specific information on 10 source categories and annual national totals are provided by CORINAIR (Table 1). The specification of the source categories into surface and elevated sources was done in consistence with the EMEP inventory. The EEM transforms the national totals to the specified model grids. For the grid transformation, the spatial information of the EMEP emission inventory is used, and additionally, a population density weighting is applied for considerably smaller grid resolutions than the original EMEP resolution. Furthermore, based on
these national totals, hourly input data of the emitted model species are calculated by the EEM considering diurnal emission profiles, changes between workdays and weekends, as well as seasonal variations for each source category.
Reformulated and Alternative Fuel Scenarios for 2005 Besides the expected benefit in air quality, realistic future emission scenarios, which promote the use of reformulated and alternative fuels, should also consider technical and economic criteria feasible inside the given time frame. Table 2 qualitatively summarizes the emission reduction potentials of reformulated gasoline (RFG) and several alternative fuels. A qualitative overview of some technical and economical criteria is given in Table 3. Diesel blends with dimethoxymethane (DMM) or dimethyl ether (DME) show a promising emission reduction potential; VOL. 36, NO. 14, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3149
FIGURE 3. Road traffic VOC emissions (kt/year) for the different scenarios divided into the VOC classes alkanes, alkenes, aromatics, and carbonyls in the N1 model domain. See Table 4 for scenario definition.
however, a massive economical availability of these additives in the near future is questionable. The overview was used as a basis for scenario definition. Only an increased use of CNG and fuel reformulation seems to be realistic, considering the time horizon of 2005 in combination with the technical and economic feasibilities. The fuel reformulation investigated in this study is basically a reduction of the fuel sulfur level and the content of aromatic compounds. A general overview of the investigated emission scenarios and simulations performed is given in Table 4. The scenarios are described in more detail in the following subsections. Trend Scenario (TREND). To take into account the future emission development based on the current legislation and technologies available, an emission forecast for the year 2005 has been performed. Two databases of emission forecasts have been considered: EMEP (15) and Prognos (16), both supplying national annual totals. The latter provides data only for member states of the European Union (EU), whereas EMEP also gives information for several other European states. However, the Prognos forecast is more specific for Germany because it also includes information on several source categories. To prepare the emissions for the domains N0 and N1, we therefore used the Prognos forecast for EU states and the EMEP forecast was incorporated for non-EU states. Using this national emission forecasts the EEM was applied to calculate the actual CTM2 emission input data sets. The composition of the fuel used can significantly influence the emitted NMVOC mixture. New data on NMVOC speciation of vehicle emission have become available recently. This includes direct vehicle exhaust measurements as well as speciated emission inventories (17-20). Therefore, the 2005 forecast simulation (TREND) was performed with updated NMVOC emission profiles for the vehicle types gasoline and diesel as well as for the contributions from evaporative losses of the source category road transport. The fleet composition follows the data given by Samaras et al. (21). The results of this scenario are used as a baseline for the other 2005 fuel scenarios. 3150
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 14, 2002
Low Sulfur Gasoline Scenario (LSG-a, LSG-b, LSG-c). The low sulfur gasoline scenarios are based on the information given in EPEFE (18) and Quissek et al. (22). In the framework of the EPEFE program, vehicle exhaust emissions of several gasoline fuels with different sulfur contents have been analyzed. Reducing the sulfur content from 380 to 18 ppm caused a reduction of 11% and 9% of the NOx and NMVOC emissions on average over the vehicle fleet investigated. Quissek et al. (22) compared the reduction potential of California Phase II gasoline (sulfur content
