Exhaust Emissions from In-Use General Aviation Aircraft (2016)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

30 Exhaust Emissions from In-Use General Aviation Aircraft of PM, when compared to using EDMS default emission indices. The Upper Limit Scenario deliberately introduces values much higher than the Average Scenario and would overestimate results in 97.5% of cases. For example, VOCs increase by 1,447% when using the upper limits as compared to the default EDMS values. These 95% confidence intervals are undesirably wide. They are also taken assuming a Gaussian distribution of emitters. As shown in Chapter 3, except for CO, those distributions are not Gaussian. The following section will investigate a more sophisticated approach toward getting confidence intervals on this sensitivity analysis. This approach aims to shrink these confidence limits and incorporate knowledge on the distribution shapes. Overall, the results of this sensitivity analysis indicate that the effect of piston engine vari- ability is much larger than the effect of updated emission indices. EDMS/AEDT aircraft/engine selections and default aircraft engine indices are representative, within our confidence intervals, of measured aircraft engine emission indices for many pollutants. This is despite seemingly large differences in total yearly emissions. The wide confidence limits highlight the need to improve and better use knowledge about the distribution of emitters, either through advanced statistical methods or with high numbers of repeat measurements of commonly used piston engines. The extensive engine mapping required to perform this analysis for a hypothetical GA airport further demonstrates the need to expand this limited database of GA aircraft and provide model users with a greater range of aircraft/engine choices. Using Monte Carlo Methods to Improve Airport Emissions Estimates The emissions at a hypothetical airport were simulated by sampling from the measured emission indices. This “Monte Carlo” approach is different from simply using the average indices for each aircraft and allows the variability of the results to come into play. The confidence limits in the final airport emissions were shrunk significantly. Monte Carlo methods are a promising tool for pinning down yearly emissions burdens, as long as the input data (e.g., time in mode, large number of emission indices) are sufficient. Despite a large variability in emission indices, it is possible to constrain an airport’s emissions tightly over time. The sample of GA engine emissions produced in this research exhibits variability. It is difficult to forecast the overall uncertainty in using the average emission indices to compute a burden at an airport; traditional assumptions about normal distributions do not necessarily apply to this limited sample of engines coupled with their variable EIs. Here, the research team describes a Monte Carlo (MC) simulation that pits the variability in the EI data against the number of operations at the hypothetical GA airport. The basis of the Monte Carlo simulation will be to draw on the pool of EI data for each engine type described in Table L-1. To compute the airport emissions burden, the annual number of operations for each engine type will be summed with each LTO by drawing a random aircraft test point. In this way, all ACRP, EDMS, ICAO and FOCA data types are sampled. Figure 4-6 illustrates the procedure. In the base case MC simulation, the variance of the fuel flow is considered to be Gaussian or normally distributed. The base times in mode are common for all operations and all engine/ airframe types (see times in Figure 4-3). The base time in mode is modified by a factor from 0.5 to 2.0, using an asymmetric distribution centered on 1, so as to induce additional variability in the emissions burden. The simulation case “EDMS” uses only the EDMS engine types and a tabulation of those engine substitutions. The simulation case “EDMS + FOCA, FAA Aircraft Engine Emissions Database (FAEED), ACRP” combines all test data in the pool of emissions data to draw from, including the fuel flow rate data.

Sensitivity Analysis on Airport Emissions 31 Figure 4-7 illustrates the differences observed in the net distributions of CO emissions per LTO. This figure is constructed by repeatedly drawing from the sample pool of emission indices until the results converge. The total CO burden to the airport is shown in grey for the baseline (left) and updated (right) scenarios. The individual contributions of the power states are shown in color. In this case, although the size of the sample pool has gone up by an order of magnitude, there is little change in the central value of the distribution (grey). This similarity means that the n = 2 CO emissions data points in AEDT/EDMS are representative of the larger n = 22 sample pool measured here. This result is expected because CO emissions are found to be Gaussian dis- tributed (see Figure 3-6). Results from the emissions species HC and NOx do not show this same Figure 4-6. Schematic representation of the Monte Carlo simulation of total emissions burden at the airport. Many of these “annual” simulations are performed to deduce the distribution of emissions. Figure 4-7. Distribution of CO emissions burden per LTO. The left hand panel is the EDMS only simulation result for numerous synthetic LTOs. The right hand panel includes EDMS with the test data from this project. The grey line is the distribution resulting from the sum of the LTO phases.

32 Exhaust Emissions from In-Use General Aviation Aircraft trend. Figure 4-7 also suggests the climb-out phase of the LTO is responsible for the greatest share of emissions. Generating distributions of emissions by LTO state (and total) was repeated for the two engine lists used in the sensitivity analysis. Employing MC simulations this way does not account for any usage profile bias—all engines in the sample pool are sampled equally, which is unre- alistic. The distribution of individual LTOs is likely dominated by professional activity at the airport (e.g., planes owned by flight schools). Nor will this MC simulation analysis account for potential ambient temperature or summer/winter fuel blend effects. The advantage of using this simulation approach is that it should empirically arrive at a distribution of emissions burdens indicative of the uncertainty in the source data. At this hypothetical airport, the large number of operations over the course of a year will tend to narrow the uncertainty in overall burden, approaching the mean value over the course of the year. For this reason, the weekly emissions were tabulated to retain some of the parent variability. Figures 4-8 through 4-10 show the results for CO, HC and NOx for the baseline EDMS and combined data scenarios. The central values of these MC results can be compared to the results of the EDMS/AEDT analysis presented earlier. The same trend in CO emissions was observed in the AEDT-based sensitivity analysis, where total annual CO emissions burdens decreased slightly. The comparison of the HC emissions trend observed in this analysis was also observed in the AEDT-based sensitivity analysis. There, the factor increase was about 3-3.3 depending on the particular class of hydrocarbons, which agrees with this alternative approach. The NOx comparison is not as good as either the CO or HC results. In that analysis, the base- line EDMS result of 2 short tons per year (st yr-1) increased 50% to 3 st yr-1. The MC analysis sees a much greater increase of 277%, when including all of the engine data. The underlying reasons for this disagreement are unclear. Figure 4-11 shows the individual contributions of different power states to the total burdens for the updated scenario. These graphs reflect the trends discussed in Chapter 3: low power states like taxi and idle contribute most to HC emissions; high-power states like take-off and climb-out Figure 4-8. CO emissions burden change between the two scenarios.

Sensitivity Analysis on Airport Emissions 33 Figure 4-9. HC emissions burden change between the two scenarios. contribute most to NOx emissions. The pie chart for CO emissions looks similar to the pie chart for fuel burn, again reflecting the observation that CO emissions are relatively constant at all powers. Although the trends in results are expected to be similar to the EDMS/AEDT approach, the MC method should yield significantly different results for the uncertainties. Figure 4-12 com- pares the emissions burdens for the two approaches. The EDMS/AEDT approach uses simple 95% confidence limits. These limits are many times larger than the converged MC uncertainties. Figure 4-10. NOx emissions burden change between the two scenarios.

34 Exhaust Emissions from In-Use General Aviation Aircraft Figure 4-11. Partitioning of emissions burden by power state. Figure 4-12. Comparison between EDMS/AEDT and Monte- Carlo estimates of a hypothetical airport’s emissions. Quantities of pollutant are reported in mega grams per year (Mg/yr). Indeed, one of the reasons why the MC results have such small confidence limits is due to the large number of operations at the hypothetical airport. Even when running a mere week’s worth of operations, the variability in the EIs for the datasets is statistically collapsed. This significant shrinking of the confidence limits is very promising. It means that, despite a large variability in emission indices, it is possible to constrain an airport’s emissions tightly over time. There are important caveats to this method, however—the test pool of emission indices must be representative of the airport’s fleet. The more measurements available, the more cer- tain that the MC method is converging to a real answer. In this case, even the large number of measurements done (e.g., 15 separate Lycoming O-320 engines) is not very big compared to the

Sensitivity Analysis on Airport Emissions 35 number of operations to be simulated (e.g., over 20,000 operations). Times in mode may also introduce a significant bias if they are considerably different from the defaults, and it would be valuable to have real operational data to verify this assumption. The results of this analysis suggest measuring additional emission indices and constructing representative distributions of emitters would be valuable. Such measurements would need to sample large numbers of aircraft (thousands, not dozens) in an automated way in the course of their normal operations. Getting a better handle on the true shapes of the emissions distribu- tions will result in more certainty in the accuracy of airport emissions estimates. Monte Carlo methods are well suited to use such distributions of emitters and assess their effect on airports.

36 C H A P T E R 5 Previous chapters have discussed the observed variability in emission indices between repeat measurements of the same engine. In doing these measurements, the research team discovered a complex landscape of conditions and variables that all affect the measured emissions from GA aircraft. Because these conditions generally get lost as the data is rolled into the averaged emission indices tables, in this chapter, the research team examines some of these parameters in detail. These conditions account for some of the variability between measurements. More important, such knowledge enhances understanding of GA aircraft and can highlight places where EDMS/AEDT-style emission inventories may gloss over details. Pilot Mindset on Fuel Mixture During the engine tests, pilots were asked to operate their engines as they usually would. Two “schools of thought” became apparent regarding fuel/air mixture in carbureted piston engines: • Full-rich at all times • Lean it out whenever possible The full-rich at all times mindset was common among many of the mechanics and pilots who operated their aircraft during the test. The mixer is pushed all the way forward, giving a maximally rich fuel/air mixture, known as “full-rich.” Advantages and disadvantages are presented below. Advantages: • Engine remains cooler at all times. Excess fuel leads to incomplete combustion, which pro- duces less heat. Overheating was of particular concern during the ground tests (where there was reduced air flow for cooling). The engine is kept cool by deliberately reducing the efficiency/ adding more fuel than needed. The theoretical engine temperature eventually begins to decrease again with extremely lean operation, and some aircraft manuals recommend operation in this region (FOCA 2007a, b) (none of the aircraft that the research team encountered). • Simple. For beginner pilots or those not used to operating a given aircraft, full-rich operation requires less fiddling back and forth between the throttle (propeller RPM) and mixer (fuel/air mixture) to achieve a stable combustion state with no cylinder misfires. • Safe. An engine running full rich will not stall. Pilots may not want to take risk stalling, even at cruise altitudes. Disadvantages: • Inefficient. A significant amount of fuel goes unburned. • High CO and HC emissions. The research team estimates that at least 8% of the potential thermodynamic efficiency is not being converted to heat to move the piston and generate work. Other Parameters Affecting Emissions

Other Parameters Affecting Emissions 37 The lean it out whenever possible mindset was usually held by pilots with newer training, and particularly pilots operating aircraft with gauges showing the exhaust gas temperature (EGT) and cylinder head temperature (CHT) for one or all of the engine cylinders. When available, EGT can be used to determine the optimal mixture for efficiency. One 1970s era aircraft had even been retrofitted with an EGT gauge and sensors for this purpose. The research team also worked with a flight school that focused on retraining pilots wishing to increase their fuel economy. Pilots of both mindsets run the engine at full rich during take-off and final approach, when the aircraft is close to the ground and stalling can be disastrous. Although the pilots may consider a given state “lean,” the true stoichiometry of the fuel mixture is likely still rich compared to the ideal mixture. Advantages: • Longer engine life. Operating with a lean mixture will prevent deposits in the engine. • Better fuel economy. Lean combustion consumes less fuel for the same power. • Lower emissions of HC and CO. More fuel is converted all the way to CO2. Disadvantages: • Requires more pilot experience. EGT and CHT gauges must be monitored, if available, and the throttle/mixer adjusted accordingly. If no such instruments are available, the pilot must be able to “feel” when the mixture is getting too lean by the sound and vibrations of the engine. • Too lean a mixture will cause misfires in the engines and the aircraft will eventually stall. • Higher NOx emissions. Higher combustion temperatures produce more NOx. • Requires that all cylinders have comparable combustion properties. If cylinders do not fol- low the same EGT temperature trends for the same mixtures, running lean on cylinder 1 could cause overheating in cylinder 6, and so on. This can decrease engine lifetime. These approaches largely determined the richness/leanness of the measured mixing states, because pilots and mechanics were, for the most part, averse to changing the mixture for concern about the risk (real or perceived) of engine health. Given that cruise is not considered in airport LTO emissions calculations and that take-off and final approach are performed at full rich by all piston engine pilots, taxi and idle emissions probably will be the most affected by these differences in aircraft operation. Most older piston engine aircraft, lacking fuel injection or EGT/CHT tech- nology, probably will be operated at or near to full-rich mixtures at all times, given the challenges and risks involved in lean operation. CO2 Carbon Fraction as an Indicator of Combustion CO2 carbon fraction is the most direct indicator of engine operation. CO2 carbon fraction describes how much fuel carbon is completely converted into CO2. In Equation 5-1, TotalC = CO2 + CO + THC, and this ratio is computed directly as a side effect of computing the emission index. A CO2 carbon fraction of 1 indicates ideal combustion where all fuel carbon is converted into CO2. CO carbon fraction CO 2 2 TotalC = Eq. 5-1 In Figure 5-1, CO2 carbon fractions for individual test points are plotted as a function of the percent of maximum achieved propeller RPM. Individual aircraft are shown as different marker types and colors, with common engines having the same label in the legend. Figure 5-1 shows the great range of measured fractions. Very few data points exceed CO2 carbon fractions of 0.8.

38 Exhaust Emissions from In-Use General Aviation Aircraft Of those data points, most belong to turbofan engines: the Garrett AiResearch, Pratt & Whitney Canada, Williams and General Electric engines. Earlier the research team showed that piston engines produce a significant amount of CO in all engine states. A measure of fuel oxidation can be devised to include both CO2 (complete oxidation of fuel carbon) and CO (“halfway” oxidation of fuel carbon). Plotting this oxidized carbon fraction vs. percent max RPM (Figure 5-2) results in higher fractions than in Figure 5-1, indicating that the cause of the low CO2 carbon fractions is the high amount of CO. With this Figure 5-1. CO2 carbon fraction vs. % maximum RPM for individual test points. Figure 5-2. A measure of carbon oxidation vs. % max RPM for individual test points.

Other Parameters Affecting Emissions 39 figure, one starts to recover a curved shape to the data, with very low and very high-powered engine states operating at generally reduced conversion efficiencies of fuel carbon to oxidized carbon and intermediate engine states operating at higher conversion efficiencies. One key difference between the emissions from aviation piston engines and turbofan engines is the ratio of carbon monoxide to carbon dioxide (CO/CO2). With the turbofan engine, aside from the near-idle engine condition, the thermodynamic efficiency is very high. With aviation piston engines, the observed carbon monoxide emission index is ~ 500 – 1200 g CO/kg fuel. Relative to the CO2 EI of 3160 g/kg, this is a more significant portion of the combustion carbon than is typi- cal of turbofan engines. (Aviation piston engines are operated differently from ground vehicle engines, with only the former having highly elevated CO emissions.) Using the combustion of CH2 as a proxy fuel, one can evaluate the loss of potential thermal energy by producing CO instead of CO2. In Table 5-1, the net loss of heat energy potential from the combustion inefficiency is bounded at ~25% on a per carbon basis. Thus, when one quanti- fies an emission index of CO of ~ 1000 g/kg fuel, this equates to ~30% of the combusted carbon not going to CO2 and the combined effect is a basic inefficiency of ~ 8%. Thermal NOx and Rich vs. Lean Combustion In urban photochemistry, the NOx species (predominantly NO and NO2) undergo transfor- mations that lead to the production of ozone. At lower altitudes, ozone is a direct health hazard and is the key component associated with the observation of modern smog. NOx compounds are only part of the system of reactions that lead to ozone production. In many regimes, the balance of various trace level chemical species is termed, “NOx limited.” This prompts the adoption of NOx control or emissions limits to mitigate degraded air quality. The ACRP Project 02-54 research sought to quantify the emissions rate using two factors: (1) fuel flow rate (commonly expressed as gallons per hour) and (2) the fuel-based emission index (determined by the observed enhancement of the pollutant in the exhaust relative to the sum of all forms that fuel carbon can take). Combustion of fossil fuel hydrocarbons in the engine pro- duces no change in number of carbon atoms, but changes the molecules in which they are found. The research team used this conservation principle in determining the emission index. The three elements that describe the flame characteristic of the combustors that are subject of this research are typically piston driven, aviation gas burning, and pre-mixed fuel to air. This is in contrast to the commercial aviation engine type, commonly a high-bypass ratio turbine engine. The emissions characterization framework, with engine states in the landing take-off cycle (LTO) exhibits large contrasts in both the fuel flow rate and the fuel-based EI values. The NOx emission indices in general aviation have relatively less sensitivity to the engine state - CH2 + CO2 Grxn(298) Sf(298) Hf(298) - +H2O +H2O Grxn(298) Sf(298) Hf(298) 93 -26.4 -57.8 -177 kcal/mole 46.3 49 45.1 47.3 -2.9 cal/mole/K =-176 kcal/mole 3/2 O2 93 -94 -57.8 -244.8 kcal/mole 46.3 49 45.1 47.3 -23.6 cal/mole/K = -237.7 kcal/mole CH2 + O2 CO-> -> Table 5-1. Comparative loss of thermal potential from combusting to CO instead of CO2.

40 Exhaust Emissions from In-Use General Aviation Aircraft (or named LTO mode) and greater dependence on the air-to-fuel ratio (or rich vs. lean com- bustion). Although this is a broad generalization, the contrast is drawn from the differences seen between piston motors and turbofans. This section illustrates this point by examining a measurement of the NOx emissions at a named engine state at two different fuel-to-air ratios. However, the basis for this comparison requires a brief discussion of the actual combustion properties that lead to the production of thermal NOx during combustion. Combusting fossil fuel produces heat and gas expansion in the piston manifold to produce work. Ideally, combustion runs to completion and the only products are CO2 and H2O. This is shown in the chemical reaction sequence below, where the consumed number of molecular oxygen (n) is dictated by the carbon-to-hydrogen ratio in the fuel (expressed here as x and y). C H n O xCO yH Ox 2y 2 2 2+ →→→ + The idealized chemical schematic holds that the yields of CO2 and H2O are matched to the fuel content of carbon and hydrogen. In reality, combustion does not match the ideal reaction scheme because it does not account for other forms of fuel carbon. When combustion does not proceed to the ideal completion, the stable product carbon monoxide, CO, is created. As a result, a useful metric for evaluating the combustion efficiency of the system is the CO/CO2 ratio observed in the exhaust gas. The less-ideal expression is depicted in the chemical scheme below that now accounts for some combustion efficiency “slip” of ideal CO2 emerging as CO and HC (sum of all other fuel hydrocarbons). a b cC H n O CO CO HC y H Ox 2y 2 2 2+ ′ →→ + + Σ + ′ NOx is functionally defined as the sum of reactive oxides of nitrogen, which in the context of emissions is adequately approximated as the sum of NO and NO2. Molecular oxygen is the needed oxidant to carry the combustion, but it is only present by volume in air at ~20%. Most of the air volume is composed of molecular nitrogen (N2). NO is produced at high tempera- ture combustion when there is enough energy to break the strong triple bond present in N2 (Zeldovitch 1946). There is very little organic nitrogen in the fossil fuel source and the effective combustion temperature and time spent at high temperature are the drivers for producing NOx. During the engine test depicted in Figure 5-3, the pilot operated the engine according to the test protocol. The test sequence was defined by the canonical named engine states (e.g., idle, taxi, take-off). The final test condition, however, involved a repeat of the engine revolution speed and fuel flow rate associated with cruise, but with additional air added to the combustion mixture. Leaning the fuel mixture is anecdotally known to result in greater EGT. This is the symptom of elevated combustion temperature in the various pistons. In Figure 5-3, the CO/CO2 observed in the lean-cruise test is lower than in the rich-cruise (the black and red tracers are on top of each other in the rich-cruise, but black, CO, is lower than red, CO2, in the lean-cruise). This suggests that the combustion is proceeding more efficiently. With the increase in efficiency, the temperature is also increasing, thereby producing more NOx. Transient Emissions Are Negligible In this research, a transient of a GA engine is defined as a deliberately included variation of engine operation from one steady-state condition (e.g., idle or approach) to another. The accu- rate quantification of the steady-state conditions of engine operation and emissions is necessary

Other Parameters Affecting Emissions 41 for the investigation on GA emissions. The aviation industry has recognized that the operating conditions within each component also need to be understood when moving between steady- state conditions. The determination of gas and PM emissions at engine transient conditions may involve different methods of instrumentation and analysis from the steady-state measurements. During field measurements, the research team observed engine transient events as the pilots were adjusting engine operating parameters such as fuel/air ratio and engine throttle. Inspection of the time series in such engine tests reveals that both NOx and THC emissions are sensitive to transient events in the engine. This section aims to evaluate the sensitivity of the total LTO burden to transient emissions. A sample aircraft (Unique ID 6, Continental O-200-A) was examined to assess the magni- tude and effect of transient emissions. The first and most important transient observed was the startup transient during engine start (Figure 5-4). The transient was most evident as an increase in PM emissions (e.g., APC and EEPS traces, pink and brown) along with a slight increase in HC compared to the steady-state idle condition. The effect of startup emissions on commercial aviation was investigated in the research that resulted in ACRP Report 63 (Herndon et al. 2012). Other transients were observed for this aircraft. A transient from take-off to idle showed emissions of NOx dropping significantly during the transient, with slight increases in acetylene and other hydrocarbons (including HC). A transient from approach to idle, on the other hand, showed increases in EI for all species, except NOx. These and other transient observations show that the exact nature of transient emissions is highly dependent on initial conditions. To assess the burden of these transient emissions, LTO emissions were calculated. Figure 5-5 shows two LTO emissions burdens graphs, where the colored area under the rectangles indicates the relative contributions of each state to emissions. Times in mode were chosen as in Table 3-1 with each transient lasting 10 seconds, except for the startup transient, which lasted 30 seconds. The left-most graph shows emissions without transients; the right-most graph shows emissions Figure 5-3. NOx and combustion carbon time series. The upper panel depicts the total NOx and specific NO2 in parts per billion by volume during an engine test. The lower panel charts the matching time series of CO2 and CO. Three particular engines states are identified in the pastel time periods: rich-cruise, take-off and lean-cruise.

42 Exhaust Emissions from In-Use General Aviation Aircraft 3500 3000 2500 2000 C H 4 [p pb ] 18:28:30 10/7/14 18:28:45 18:29:00 18:29:15 18:29:45 18:30:00 18:30:15 Time [UTC] 50 40 30 20 10 C 2H 6 [p pb ] 1600 1200 800 400 0 C 2H 2 [p pb ] 300 200 100 0 C 2H 4 [p pb ] 430 420 410 400 390 C O 2 [p pm v]50x10 3 40 30 20 10 0 C O [p pb ] 15 10 5 0 N O x [p pb ] 8 4 0 N O [p pb ] 14 12 10 8 6 4 T H C F ID [p pb ] 8 4 0 S O 2 [p pb ] 250x10 3 0 A P C n um [# c m -3 ] 800x10 3 400 0 E E P S n um [# c m -3 ]20 10 0 M A A P [u g m -3 ] 0.8 0.4 E E P S m as s [u g m -3 ] idleStartup Transient 18:29:30 Figure 5-4. Changes in emission signatures between the startup transient and the idle state. Figure 5-5. Emissions burdens for a sample LTO cycle with (right) and without (left) the effect of transient emissions.

Other Parameters Affecting Emissions 43 with the addition of transients. Examination of these figures shows that the effect of transient emissions on the LTO burden is negligible. In fact, the duration of the transients were so short that only the startup transient is visible. Table 5-2 summarizes the magnitude of these transient emissions and compares them to the total burden without transients. The contribution to the total is negligible, only 0.3% of the totals at most. Ambient Conditions The research team investigated the influence of ambient temperature variation on HC emis- sions by plotting emission indices versus temperature. Measurements were made in spring and autumn to limit the effect of temperature variations. The lowest ambient temperature observed during a piston engine test was 283.6 °K and the highest was 297.0 °K; the ambient barometric pressure was between 100.4 and 102.7 kPa. Within the small range of barometric pressures, no effect on EI was observed. Ambient temperature varied somewhat more widely. Figure 5-6 presents correlation plots of CO and HC emission indices from measured piston aircraft engines vs. ambient temperature. Red markers correspond to measurements at idle condition and blue markers to measurements at T/O. A linear fit was performed to the data, and the slope (m), intercept (b), and coefficient of correlation (R2) are reported. Note the poor correlation observed for all cases, with coeffi- cients of correlation of R2 < 0.18; a “good” correlation would have an R2 > 0.75. These R2 values mean that temperature effects are not important within the observed temperature range. This is due to the dual effect of variability in emission indices and the limited temperature range Increased emissions due to transients HC CO NOx nvPMm (MAAP) g/LTO g/LTO g/LTO g/LTO N7482G Connental O-200-A 55 4735 7.43 1.391 +1 +14 +0.01 +0.006 2 Table 5-2. Impact on LTO burden of aircraft N7482G from transient emissions. Figure 5-6. Correlation between emission index and temperature for piston engine CO and HC emissions. The slope (m), intercept (b) and coefficient of correlation (R2) are shown for each fit. Idle and T/O power states are shown.

44 Exhaust Emissions from In-Use General Aviation Aircraft sampled. A change in emission indices for measurements taken at more extreme temperatures or barometric pressures would be expected. Previous studies also investigated the effect of ambient temperature. Appendix 3 of the FOCA report (FOCA 2007b) states that hotter ambient temperatures result in richer running engines. Though correlation is very poor for the research team’s results, this is consistent with the data in Figure 5-6. Furthermore, FOCA recommends using temperature effects of 0.016 g HC per kg fuel per K and 3.1 g CO per kg fuel per K to correct emission indices taken at non-ambient temperatures (FOCA 2007c). These FOCA values are taken from data span- ning a 30 Kelvin temperature difference, much larger than the ~13 degree difference sampled here. Studies have also been done on the effect of ambient conditions on jet engine emissions, both for PM (Gleitsmann and Zellner 1998) and for HC emissions (Herndon et al. 2012). Fuel Additives Numerous aircraft tested exhibited unusually high toluene emissions and their fuel also revealed elevated levels of toluene compared to fuel samples pulled from fixed-base operator (FBO) supplies. Figure 5-7 compares the burden of HC per LTO cycle to the burden of toluene only. Points are marked with the unique ID of the aircraft. Several high-toluene aircraft populate the upper part of the graph (e.g., aircraft 26) while most aircraft are clustered at modest toluene burden levels. The presence of toluene does not correlate to higher total HC burdens. The observation of toluene in both fuel and exhaust suggests that the aircraft owners were using a non-traditional fuel such as premium unleaded (Mogas) or a commercially available fuel additive. Mogas can be eliminated because it contains a mixture of aromatic compounds (i.e., benzene, toluene, xylene). The research team believe the source of the toluene can be traced to a fuel additive called Alcor TCP. TCP or tricresyl phosphate is used to help alleviate the condensa- tion (fouling) of lead onto the spark plugs and valves of older low-compression piston aircraft engines. The only commercial AVGAS is 100 Low Lead, which uses tetraethyl lead to achieve its octane rating. Combustion of tetraethyl lead produces lead oxide, which is not volatile at normal 5 67 8 9 10 1112.112.214 15 178 19 2024 25 26 27 30 3132.13334 35 36 37 38 3941 42 43 444647.2 49 54 57 25 20 15 10 5 0 T ol ue ne B ur de n [g T ol ue ne /L T O ] 12008004000 HC Burden [g HC as CH4/LTO] Figure 5-7. Toluene emissions burden per LTO compared to hydrocarbons emissions burden per LTO for measured aircraft.