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.

Trends in Emission Indices 15 EI distributions change with increasing power (idle > taxi > cruise > take-off) as expected. HC emissions decrease with power. CO emissions largely stay the same. NOx emissions increase with power, peak at cruise, and then fall again. PM distributions behave differently depending on their measure (count vs. mass) and on their composition (total vs. non-volatile only). This section explores the variability and trends in observed in carbon monoxide (CO), hydro- carbons (HC), nitrogen oxides (NOx), and PM emissions from GA piston engines. Figure 3-5 shows the expected behavior of the gaseous exhaust species as a function of the oxygen left in the exhaust, which scales inversely with the fuel/air (F/A) ratio. At the vertical 0 line, one achieves ideal stoichiometric combustion where all fuel is combusted to CO2, and all oxygen is consumed. Left of this line, the engine is operating with excess fuel (rich) and produces high amounts of CO and HC and low NOx. Right of this line, the engine is operating with excess air (lean) and pro- duces high NOx but low CO and HC. The grey arrow in Figure 3-5 shows the expected behavior of the expected fuel/air ratio as a function of power state; for the piston engines examined here, one would not expect lean of stoichiometric behavior (hashed area). Although take-off (T/O) is the highest power state, it is not the leanest state; cruise is generally the leanest power state mea- sured. (The reasons behind this are explored in “Pilot Mindset on Fuel Mixture” in Chapter 5.) The emissions trends suggested by Figure 3-5 are borne out in the data from the piston engines. In Figure 3-6, aircraft count is plotted against emission index bins to yield distribu- tions of EIs for measured piston engines. Four selected power states were chosen: idle, taxi, cruise and take-off. The approach, final approach, and climb-out states resembled the take- off state and were omitted for clarity. All EI axes were logarithmically scaled, except for CO, to highlight the orders-of-magnitude differences observed among emissions from different aircraft. Grey arrows show how the peak emissions distributions change with increasing power state. As shown in Figure 3-6, except for CO emission index, lognormal distributions were observed for HC, NOx, and PM emissions. Lognormal distributions are skewed distributions: the most common emission index (the mode) is smaller—sometimes much smaller—than the average emission index. These distributions demonstrate orders-of-magnitude difference in emissions among piston engines, even of the same engine model. Given that a lognormal distribution often results from a statistical multiplicative product of several independent variables, the research Figure 3-5. Expected exhaust gas composition as a function of the richness of combustion.

16 Exhaust Emissions from In-Use General Aviation Aircraft team’s observation of lognormal distributions indicates that many random variables (e.g., fuel- to-air ratio (F/A), ambient temperature, and pilot preference of aircraft operation) are involved in determining the HC, NOx, and PM emissions from piston aircraft engines. In contrast to other species, CO emissions follow a normal distribution. The CO emissions are relatively constant with engine power, except at cruise condition, where CO emissions flat- ten out with the lowest mean value. Given that CO is an indicator of incomplete combustion Figure 3-6. Distributions of emission indices as a function of power state (idle, taxi, cruise, take-off). Logarithmic axis for all EIs except carbon monoxide. The grey arrows show how the peak of the distribution moves with increasing power.

Trends in Emission Indices 17 (see “CO2 Carbon Fraction as an Indicator of Combustion” in Chapter 5), formation of CO is sensitive to the internal combustion engine temperature and F/A ratio. At fuel-rich combustion conditions, CO concentration normally increases with F/A ratio. When the F/A ratio closes to the stoichiometric condition, CO starts to decrease dramatically. As shown in Figure 3-6, CO emis- sion increases from idle to taxi, to take-off, and then to cruise, indicative of a decreasing F/A. The research team’s CO emission results reflect that pilots prefer to operate rich in all engine states except for cruise (see “Pilot Mindset on Fuel Mixture” and “CO2 Carbon Fraction as an Indicator of Combustion” in Chapter 5). CO emissions at this lean-cruise condition are, on average, almost half those in the T/O or C/O conditions (rich). The lognormal distributions in Figure 3-6 indicate that HC emissions decrease with increasing engine power. HC emissions from piston engines mainly consist of unburned and slightly burned fuel (see “Hydrocarbon Emissions from GA Are Primarily Unburned Fuel” in Chapter 5). Three main processes are expected to contribute: 1. Engine misfire, 2. Wall quenching, and 3. Combustion chamber deposits. Additionally, HC concentrations can be influenced by the temperature of the fuel-to-air (F/A) mixture as it enters the combustion chamber, so large changes in ambient temperature could have an effect. The effect of ambient temperature on piston engine measurements is explored in Chapter 5. The observed lognormal distributions for HC emissions are consistent with these multiple influential variables. As demonstrated in Figure 3-6, NOx emissions from piston engines are inversely correlated with F/A and lognormally distributed. In this research, NOx emissions are highest at cruise, decreasing in the following order: cruise → take-off → taxi → idle. This order implies that F/A at cruise is the lowest and is the highest at idle—this is in agreement with the observations from CO and HC emission measurements. Shapes of the PM emission distributions are broader and harder to define than those for HC and NOx emissions. In addition, the skewness, or asymmetry, of the distributions also becomes much larger, especially for the nvPMm emissions. The broad distribution and large skewness imply that additional measurements and analysis are necessary to understand the source and evolution of PM emissions from piston aircraft engines. For piston aircraft engines, measurement results indicate black carbon soot emissions (nvPMm) are larger at the low-power conditions (idle and taxi), contrary to the observation from turbofan aircraft engines. The total soot emissions at a GA airport will be dominated by a few high emitters. At take-off, for example, the largest three emitters contribute 50% to the total emissions from 44 aircraft engines. Emissions of nvPMn, which include contributions from both black carbon soot and PbBr2 particles, are much less sensitive to power condition, compared to nvPMm emissions. Figure 3-6 shows a slight increase of nvPMn emission with engine power. This observation implies that PbBr2 particles are significant contributors to the nvPMn emissions, since PbBr2 emission is independent of engine power. Further investigation is necessary to distinguish the contributions from PbBr2 and black carbon soot. In general, total PM number emissions (tPMn) are ten times larger than the nvPMn emis- sions, indicative of predominance of volatile PM over non-volatile PM. Given the high level of incomplete combustion for piston aircraft engines, this observation is understandable. However, the research team also observed bimodal lognormal distributions for the tPMn emissions at each engine condition. The difference between the two modes is more than one order of magnitude.

18 Exhaust Emissions from In-Use General Aviation Aircraft The smaller of these two modes may come from nucleation mode particles generated from vola- tile material/unburned fuel, whereas the larger particles can come from black carbon soot. The contribution of lead particles to these modes is still under investigation. GA Emission Indices Show a Great Deal of Variability A major finding from this research is that emissions from GA piston engines show a great deal of inherent variability. Piston aircraft are operated somewhat by “feel.” For example, in the idle state, pilots reduce the throttle (and therefore the engine RPM) until the engine starts to run too roughly. The pilot also has direct control over the fuel/air mixture, and the research team has seen evidence of fuel additives (see following sections). Thus, for piston aircraft, the parameters that define a valid idle (or any other state) span a large multidimensional parameter space, par- ticularly when compared to a turbojet engine, where mixture is handled automatically, and the pilot can dial in a percent power for each state. The research team’s recommendation for dealing with this variability is to understand that any airport emissions inventory produced for GA will carry uncertainty bars directly related to the nature of piston engines and their operation. Policies, such as encouraging pilots to run lean (less excess fuel), could be investigated, especially during taxi and idle where there is no safety issue with stalling the engine. This is one way of mitigat- ing airport emissions of hydrocarbons and CO, but with a potential increase in NOx. Quantitative Validation of Existing Data The inherent variability in the data means that few existing data points can be invalidated with certainty. The most important invalid data point is the hydrocarbon emission index from the Lycom- ing O-320 engine, which is underestimated by a factor of 2.3 versus the results from this study. The research team investigated three sources of existing aircraft emissions data: 1. The Swiss Federal Office of Civil Aviation (FOCA 2007a), which includes a. Original measurements b. Data from FAA’s Aircraft Engine Emissions Database (FAEED) 2. FAA’s Emission and Dispersion Modeling System (EDMS), which will be the same data used in the new standard, the AEDT, and which includes a. Data from the Environmental Protection Agency’s Compilation of Air Pollutant Emission Factors – Mobile Sources (AP-42) (EPA 1989) b. Data from jet engine manufacturers like Pratt & Whitney 3. The International Civil Aviation Organization Database (ICAO 2013), which includes data from commercial jet engines To quantitatively validate (or invalidate) this data, the research team considered the emis- sions burden of a given engine type for a standard landing-take-off cycle (LTO). This burden, expressed as g/LTO, rolls up the emissions factors and the fuel flows for all engine states of interest. Calculating LTO burdens allowed the research team to turn a 28-dimensional problem (seven engine states multiplied by four emission species) into a 4-dimensional problem. Vali- dation was done for engines that the research team measured several times. These repeat mea- surements allowed the research team to determine with confidence the true variability between different instances of the same engine. The research team thus reports 95% confidence intervals on the average measured emissions burden. Four emission types were compared: hydrocarbons (HC), carbon monoxide (CO), oxides of nitrogen (NOx), and non-volatile PM mass (nvPMm)

Trends in Emission Indices 19 measured via engine exhaust particle sizer (EEPS). Table 3-1 lists the LTO times used. For the existing data, which does not differentiate between idle and taxi, nor between approach and final approach, the sum of the relevant times was used. Different characterization technologies were used to compare particulate quantification. The FOCA data was collected using a combination of the scanning mobility particle sizer (SMPS) and the EEPS 3090. Table 3-2 shows selected experimental data used in the comparisons. Color bars guide the eye to the magnitude of the emission burden. Variability confidence intervals are expressed as a percentage of the average so that they can be compared on equal footing: = σ  % 95%CI T avg DF where %CI is the percent confidence intervals, avg is the average of the replicate determinations of emissions burden per LTO, s is the standard deviation of these replicates, and TDF95% is the student’s T at 95% confidence for degrees of freedom (DF = count - 1). In these results, engine subtypes are neglected because no repeat measurements of any of the particular subtypes of FOCA data were acquired, and EDMS does not differentiate among subtypes. Condion T/O C/O Cruise App Final App Taxi Idle Secs (Tot) 42 132 0 210 30 660 900 Table 3-1. LTO time-in-modes used in calculating emissions burdens. Engine states considered are take-off (T/O), climb-out (C/O), cruise, approach (App), final approach (Final App), taxi and idle. Table 3-2. Experimental data for use in validation. The size of the color bars is proportional to the magnitude of the emissions burden for HC (orange), CO (pink), NOx (green) and tPMm (blue). Engine Full Tests HC Avg Variability CO Avg Variability NOxAvg Variability g/LTO % at 95%Conf g/LTO % at 95% Conf g/LTO % at 95% Conf tPMm Avg Variability g/LTO % at 95%Conf Full Engine General Electric CF34 3A1 1 292 7315 1278 7.04 Engine Family Count Lycoming O 320 16 258 38% 4083 47% 32 246% Lycoming IO 360 4 598 116% 4387 47% 44 434% 0.90 120% 2.04 358% Lycoming O 360 6 406 95% 4924 58% 16 220% Lycoming IO 520 1 968 6960 13 1.68 186% 1.95 TCM O 470 1 391 3441 11 Lycoming O 540 3 747 236% 6457 108% 21 32% Lycoming IO 540 4 795 115% 8483 96% 39 212% 1.02 3.06 444% 3.33 230% Horse Power Family diverse Prop 200hp 35 346 112% 4056 51% 26 255% diverse Prop 300hp 10 753 95% 7078 79% 27 171% diverse Prop 160hp 25 275 75% 3841 52% 25 256% 1.27 169% 2.83 188% 1.00 123%