Summarizing and Interpreting Aircraft Gaseous and Particulate Emissions Data (2008)

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.

41 D.1 PM Data Review Summary D.1.1 ICAO Smoke Number Database The ICAO Smoke Number Database (ICAO 2008) does not provide adequate quantitative parameterization of PM char- acteristics suitable for local air quality applications. The ICAO smoke number was developed as a metric for plume visibility and is measured for all engines in the commercial fleet. It is not a fundamental PM characterization parameter and can only be applied to dispersion models if correlated with fun- damental PM characterization parameters such as number, size, and mass. Attempts at such correlations have met with limited success (Paladino 1997; Whitefield et al. 2001; Soci- ety of Automotive Engineers 2004; Wayson et al. 2006; FAA Jul 2007). D.1.2 Essential Fundamental PM Characterization Parameters The essential fundamental PM characterization parameters required for local air quality applications are the following: • Number (number-based Emission Index, EIn); • Size distribution; • Mass (mass-based Emission Index, EIm); • Composition. D.1.3 Confidence in Fundamental Parameter Measurement When the APEX series of campaigns was initiated, mea- suring fundamental properties of aircraft gas turbine PM was a new challenge. Since no comparable data existed on aircraft PM, the experiments contained two separate internal checks to ensure data accuracy: (1) measurement redundancy and (2) gas-phase measurements. State-of-the-art PM mea- surement instruments were deployed during all of the APEX campaigns, and new instrumentation developments were in- corporated in the measurement suite as the campaigns evolved. Particle size distributions were measured with no fewer than three different instruments (DMS500, SMPS, and EEPS); par- ticle count was measured to determine EIn using several dif- ferent model condensation particle counter instruments (CPC); EIm was measured using real-time instruments (DMS500, SMPS, TEOM) and filter samples: black carbon soot mass was measured using two different types of instruments (MAAP and PSAP); particle composition was measured by two real- time instruments (AMS and PAS—for polycyclic aromatic hydrocarbon content) and complemented by filter samples. Additionally, the measurement suites deployed by the various teams intentionally overlapped so that multiple measure- ments of the same PM characteristics could be made in par- allel. The built-in redundancy helped ensure that operator errors could be identified and removed from the data set. Gas-phase data were used to build additional confidence in the measurement approach and in the maintenance state of the gas turbines tested. Any given engine emission can be considered representative if its measured primary combus- tion gas profiles (NOX, CO, and HC) match those calculated from certification data in the ICAO databank. The gaseous emissions of the APEX engines exhibited the anticipated trends with respect to engine power condition and the mea- sured values were similar to the ICAO certification values (see Section D.2 for more details on the gas-phase measure- ments). Therefore, the extractive sampling dilution system was judged to be operating properly and the condition of the APEX engines was judged to be characteristic of typical in-use engines representative of the fleet. D.1.4 Data Reproducibility During the campaigns, PM emissions data were collected at over 1,200 stable conditions (i.e., the power setting/fuel flow was stable at the desired set point). Table D1.1 lists all of A P P E N D I X D Additional Supporting Material for Chapter 5: Review of the Data from Measurement Campaigns

the PM species measured during the APEX campaigns, along with the instruments used to make the measurements. State- of-the-art instruments were used for the measurements, and the quality of the instruments is reflected by their fast time re- sponses and low detection limits. During an emissions test, the airplanes remained grounded and chocked during all tests while the engine thrust was varied to simulate operation at ground idle (4%), idle (7%), taxi (30%), climb-out (85%), take-off (93%), and intermediate power conditions including 15%, 45%, and 65% rated thrust. The power cycle during a typical experiment was as follows: (1) the engines were allowed to warm up for roughly 5 to 10 min; (2) measurements commenced as the engines were operated at ground idle; (3) the test continued as the power was in- creased in a step-wise fashion (e.g., 4% to 7% to 15%, etc.) up to either take-off power or climb-out power; (4) the power was directly reduced to either idle or ground idle; (5) after several minutes at idle, the power was increased directly to either take-off or climb-out; (6) the test concluded as the power was reduced step-wise back to ground idle. PM samples were taken continuously throughout the entire experiment and each stable point lasted for 2 to 5 min. Sam- pling was performed both at 1 m [3 ft] from the engine exit plane and further downstream of the engine (15 m, 30 m, 45 m, or 50 m [49 ft, 98 ft, 148 ft, or 164 ft] depending on the size of the engine). During each engine test, EI measurements were made at a given thrust rating both as the engine thrust was increased and as it was decreased to the set point. Table D1.2 lists EI number and EI mass data for all of the engines studied in the APEX series of campaigns. Idle/taxi (either 7% or 8% depending on the engine), approach (30%), and climb-out (85%) power conditions are emphasized in Table D1.2 since these are the set points used in ICAO certi- fication data. EIs for take-off power conditions are only available in certain cases due to the difficulty in operating stationary aircraft at full- rated thrust. Each EI reported in Table D1.2 is the average of all available replicate points taken at a given set of conditions for measurements made at 1 m [3 ft]. (For a given airframe/engine combination, duplicate measurements are those made at the same downstream distance and engine power condition.) Accu- rate estimation of the measurement uncertainty is critical to the proper use of the APEX data set. Typically, each EI reported is the average of between 3 and 6 replicate measurements. Experimental uncertainties were estimated by setting the un- certainty in each EI equal to the standard deviation of all available EI data points at a given set of conditions. These data handling procedures ensure that reported errors accurately represent experimental reproducibility. Sources of systematic error are considered later in this section. D.1.5 Measurement Reliability and Sources of Systematic Error for PM In addition to reproducibility, absolute measurement accu- racy is also important. Two measurement uncertainties con- tribute to overall uncertainty in the EIs determined in these studies. They are related to the detection limits and the sys- tematic errors associated with the measurement of (1) the CO2 concentration, and (2) the PM differential concentra- tion. Systematic errors for PM EIs arise from particle line loss, sample dilution, flow rate, and particle density measurements. A detailed discussion for PM error analysis, using electric mobility methodology, can be found in Schmid et al. 2004, where derived relative uncertainties (%) for EIn and EIm are 20% and 30%, respectively. D.1.6 Sample Sources These studies focused on engine specific emissions and their downstream evolution. • Dedicated aircraft – Close to exit plane (≤1 m [≤3 ft]); – Near field plume (~10 m, ~15 m, ~30 m, and ~50 m [~33 ft, ~49 ft, ~98 ft, and ~164 ft]). 42 Instrument Species Detected Detection Limit Sampling Frequency Campaign(s) Deployed Combustion DMS500 Size distribution 5 nm 1 sec APEX1,2,3, Delta-ATL TSI SMPS Size distribution 15nm 30 sec APEX1,2,3, Delta-ATL TSI CNC Total particle concentration 7nm 1 sec APEX1,2,3, Delta-ATL Aerodyne Aerosol Mass Spectrometer Volatile PM composition and size distribution (sulfate and organic) >30 nm >100 ng m -3 >3 sec APEX1, 2, 3 Thermo MAAP Black carbon soot mass >5 µg m -3 1 sec JETS APEX2, APEX3 Table D1.1. PM instruments deployed in APEX missions.

• Advected plumes from aircraft operating under normal landing and take-off (LTO) conditions – Taxi; – Take off; – Approach. To address the impact of PM emission on local air quality it is necessary to obtain both exit plane and downstream PM emission data. The emission products at the exit plane depend exclusively on the engine design and operating conditions. They evolve in the downstream plume. This evolution is greatly influenced by atmospheric conditions. The PM ob- served in the downstream plume is a complex mixture of the emissions from the engine, the results of plume processing, and the background ambient PM. D.1.7 General PM Emission Trends When sampling at or close to the exit plane (within 1 m [3 ft]), emitted particles were log-normally distributed within a single size mode and ranged from a few nanometers (nm) to 300 nm in diameter (Figure D1.1). D.1.7.1 PM Characteristics Change with Engine Operating Conditions for a Given Engine Type D.1.7.1.1 All engines. At ≤1 m [≤3 ft], i. Particle mass and black carbon emission indices (EIm and EIm-soot respectively) were a minimum at low powers 43 Engine Model/Tail Number Engine Location EIn (10 15 particles/kg fuel burned)* EI m (g/kg fuel burned)* 7% 30% 85% 7% 30% 85% CFM56-2C1 / N817NA stbd 0.44±0.146 0.45±0.198 1.92±0.367 0.0048±0.0035 0.0067±0.00340 0.13±0.0370 CFM56-3B1 / N353SW stbd 1.08±.0.565 1.17±0.499 4.20 0.0043±0.00086 0.0060±0.00003 0.254 port 1.20±0.58 1.09±0.358 0.0045±0.00085 0.0060±0.00096 No Data CFM56-3B2 / N695SW stbd 1.15±0.784 1.64±0.323 2.57 0.0062±0.00211 0.016±0.00111 0.249 port 40.1±23.8 35.9±49.6 0.054±0.0186 0.021±0.0208 No Data CFM56-7B22 / N435WN stbd 0.50±0.104 0.50±0.173 1.12 0.0090±0.00298 0.0079±0.000923 0.0614 port 0.51±0.147 0.38±0.120 0.0083±0.0033 0.0055±0.000829 No Data CFM56-7B22 / N429WN stbd 0.28±0.184 0.26±0.249 1.09 0.0021±0.00171 0.0023±0.00180 0.073 port 0.065 0.098 0.00046 0.00098 No Data JT8D-219 / 908DL stbd 2.145±1.47 0.85±0.43 11.2±0.32 0.0042±0.00298 0.0014±0.000315 0.22±0.0395 JT8D-219 / 918DL stbd 8.81±1.78 0.58±0.02 10±0.69 0.042±0.0174 0.0013±0.000133 0.18±0.0106 CFM56-3B1 / N14324 stbd 0.18±0.0977 0.20±0.0919 1.44±0.120 0.0033±0.00099 9 0.0057±0.00172 0.13±0.00990 CFM56-3B1 / N70330 stbd 0.39 0.25 1.16 0.0063 0.0042 0.0837 RB211-535E4-B / N75853 stbd 0.34 1.26±0.145 1.48 0.013 0.072±0.0112 0.475 RB211-535E4-B / N74856 stbd 0.38±0.202 0.66 1.31±0.0665 0.013±0.0056 0.035 0.36±0.0177 PW4158 / N729FD stbd 10.5±21.89 6.79±16.234 1.89±0.316 0.27±0.487 0.048±0.102 0.16±0.0122 AE3007-A1E / N11193 stbd 3.39±3.23 0.62±0.0988 0.68±0.0299 0.059±0.0631 0.016±0.00170 0.043±0.00150 port 1.36±0.238 0.68±0.0436 1.03±0.0609 0.029±0.00206 0.016±0.000909 0.080±0.00655 AE3007-A / N16927 stbd 0.93±0.581 0.76±0.340 0.72±0.0125 0.022±0.00493 0.020±0.00223 0.057±0.00313 CJ6108A / N616NA stbd 0.49±0.162 3.07±1.11 8.53±0.660 0.0078±0.00259 0.071±0.00257 0.298±0.0541 Note: Idle/taxi (either 7% or 8% depending on the engine), approach (30%), and climb-out (85%) power conditions are emphasized since these are the set points used in ICAO certification data. *All available 1-m data. Table D1.2. EI number and EI mass data for all of the engines studied in the APEX series of campaigns.

and increased with power, reaching values more than 0.3 g-particle/kg-fuel at power levels higher than 65%. ii. The mean particle diameter increased linearly with power, ranging from around 15 nm at idle to about 40 nm at maximum power (at engine exit). iii. Primary (non-volatile) particle EIn varied from 0.16 to 3 × 1015 particles/kg-fuel, and were greatest at idle and take-off thrust settings and a minimum at power levels corresponding to approach. iv. EIm values were nonlinearly dependent on engine power and typically less than 20 mg-particle/kg-fuel over the 4% to 65% engine power range and greater than 200 mg-particle/kg-fuel at and above 85% power level. v. The PM composition is primarily non-volatile:  Temperatures high enough to suppress formation of volatile species.  Validated by measurements at APEX1, JETS APEX2— not APEX3.  No dependence on fuel composition, specifically sul- fur, and aromatics. Fuel composition was systemati- cally investigated in APEX1. JETS APEX2 and APEX3 provide further determinations of the dependence of 1-m PM characteristics on fuel properties. For downstream locations (>30 m [>98 ft]) and advected plume data, i. Measured particle distributions typically exhibited two distinct modes, one corresponding to non-volatile particles and peaking at roughly the same diameters observed in the 1-m [3-ft] samples, and the other occupied by freshly nucleated sulfur and organic particles peaking at <12 nm. ii. At high engine powers, particle mass emissions were dominated by non-volatile PM. iii. Volatile PM number and mass concentrations are de- pendent on fuel sulfur concentrations. iv. Non-volatile particle size as well as EIn and EIm were in- dependent of fuel properties or downstream sampling distance (plume age). v. From the advected plume data, on any given day the engine-engine variability within a given class is less than 5% for mass- and number-based emission indices (see Fig- ures D1.2 and D1.3). vi. From the advected plume data, the day-to-day variabil- ity for a given engine class ranged from 10% to 30% for mass-based and 10% to 80% for number-based emission indices (see Figures D1.2 and D1.3). vii. Changes in ambient atmospheric conditions are likely to impact PM emissions. Advected plume data indicate that EIn is more sensitive to ambient conditions than EIm, consistent with EIn being dominated by volatile nucleation/growth mode particles and EIm dominated by non-volatile soot particles. Table D1.3 summarizes the daily and day-to-day changes in atmospheric conditions during this study. The observations discussed in items v to vii above provide powerful tools for assessing aircraft operations on airport 44 Base Fuel, 1m 1.0E+01 1.0E+02 1.0E+03 1.0E+04 1.0E+05 1.0E+06 1.0E+07 1.0E+08 10 100 1000 Dp (nm) dN /d lo gD p 0.12 0.34 0.58 0.82 0.88Fuel flow rate (kg/s): Figure D1.1. Typical particle size distributions. Source: Lobo, Hagen, Whitefield, and Alofs, “Physical Characterization of Aerosol Emissions from a Commercial Gas Turbine Engine.” Journal of Propulsion and Power, 2007, 23 (5), p. 922.

45 419 14 18 53 13 21 19 37 10 14 13 34 9 9 16 94 1941 34 309 9 12 0 25 50 75 100 125 150 175 200 225 250 9/27/2004 9/28/2004 9/29/2004 COMPOSITE EI n (10 15 pa rti cl es /k g fu el ) BR715 CF34 CF6-80 CFM56 JT8D PW 2037 Figure D1.2. Average number-based emission indices at take-off measured for six aircraft engine families during the Delta-Atlanta Hartsfield Study. 419 14 18 53 13 21 19 3714 13 10 34 9 916 9419 41 34 309 9 12 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 9/27/2004 9/28/2004 9/29/2004 COMPOSITE E Im (g /kg fu el) BR715 CF34 CF6-80 CFM56 JT8D PW2037 Figure D1.3. Average mass-based emission indices at take-off measured for six aircraft engine families during the Delta-Atlanta Hartsfield Study.

local air quality but were drawn from only one study (Lobo et al. 2008). Their potential value warrants further advected plume studies of this type. D.1.7.1.2 The CFM56 studies. The engine type most ex- tensively studied in these campaigns is the CFM56, which is the most prevalent engine type operating in the commercial fleet. The CFM56 common to all four campaigns and the CFM56 data provide the most insight into the engine-to- engine variability issue. Furthermore, the data provide a unique opportunity to examine the variability within sub-classes of the engine, specifically: -2C1, -3B1, and -7B22. At ≤1 m [≤3 ft] (see Figure D1.4): • For all CFM56 sub-classes, mean particle diameter increases with increasing engine power. No statistically significant difference between the particle number diameter of the PM emitted by the various CFM56 technologies is discernable due to measurement variability (i.e., real engine-to-engine variability) and measurement uncertainty. • Both the -3B1 and -7B22 engine sub-classes demonstrate a minimum number-based emission index (EIn) at ~20% power. The newer technology -7B22 engines produced fewer particles per kilogram of fuel burned than did the older -3B1 engines. Averaged across all powers, the -7B22 engines ex- hibited a 79 ± 12% reduction in number-based emissions normalized to fuel flow relative to the -3B1. EIn for the -2C1 engine fall between those of the -3B1 and -7B22 series. • The mass-based emission index (EIm) increased with in- creasing power. The trend is stronger for the older engine technology (-3B1). At 85% power, EIm for the 7B22 engines is 72% less than the -3B1 engines. A statistically significant decrease is also observed for the CFM56-2C1 engine com- pared to the -3B1. At ≥30 m [≥98 ft]: The onset of gas-to-particle conversion was apparent at downstream locations for low to medium powers as the formation of new particles. At high powers, gas-to-particle conversion resulted in formation of a coating on the soot particles. In the ≥30-m [≥98-ft] data, non-lognormal size dis- tributions were often observed. The ≥30-m number-based emission indices were 10 to 100 times greater than those measured at 1 m [3 ft] due to the appearance of a large quan- tity of particles smaller than 15 nm in diameter. The <15 nm particles present in the ≥30-m [≥98-ft] samples were attributed to nucleation of condensable materials as the hot exhaust gases cooled. Since the <15 nm particles do not contribute significant mass, EIm and related parameters did not vary strongly with sampling position. Because the coating layer was thin (<1 nm), the volatile coatings on the soot particle did not alter the soot particle size distribution. D.1.7.2 PM Characteristics Also Change between Engine Types At both the ~1-m and ≥30-m (~3-ft and ≥98-ft) locations: The mean particle diameter (Figure D1.5), number-based emission index (Figure D1.6), and mass-based emission index (Figure D1.7), depend on engine type for all PM parameters measured as follows: • All engines have comparable minima and • Significant variation is observed between the maxima. D.1.7.3 PM Characteristics Change during Engine Warm Up The number-based emission indices, EIn, for a given operat- ing power were observed to decrease with engine on-time following a cold start. For example, in the case of a CFM56- 7B22 engine sampled at 50 m [164 ft], EIn decreased by about 60% after the engine warmed up. The effect was observed for other engine types where cold starts were studied. D.1.8 Chemical Composition of Aviation Particles As a complement to aviation particle physical characteri- zation, the APEX studies included measurements of particle 46 Date Ambient condition min max avg std dev % Deviation 9/27/2004 Temperature (°C) 19.1 20.4 19.5 0.2 1.1% Rel. Humidity (%) 91.7 100.0 99.0 2.0 2.0% 9/28/2004 Temperature (°C) 21.5 27.5 24.8 1.3 5.2% Rel. Humidity (%) 53.0 76.9 64.8 6.3 9.7% 9/29/2004 Temperature (°C) 19.9 28.5 24.2 2.2 8.9% Rel. Humidity (%) 34.6 78.1 54.7 12.9 23.6% Table D1.3. Summary of the daily and day-to-day changes in atmospheric conditions during the Delta Atlanta-Hartsfield Advected Plume Study.

chemical composition. Just like particle mass, number, and size, the chemical composition is likely to play an important role in potential human health and environmental impacts. The aerosol mass spectrometer (AMS) was one of the instru- ments used to measure particle chemical composition during APEX. The AMS is a powerful instrument that obtains size- resolved particle chemical data. The operation of the AMS has been described elsewhere in detail (Jayne et al. 2000; Canagaratna et al. 2007; Drewnick et al. 2005; DeCarlo et al. 2006). Briefly, a specially designed inlet focuses particles in a sample gas at the expense of the gas-phase molecules. The focused particles enter a high vacuum chamber and are accel- erated to their vacuum terminal velocity. Since larger particles travel more slowly than smaller particles, they become size separated during their travel through the vacuum chamber. At the end of their travel, the particles strike a heater which is usually held at 600°C. Volatile components become vaporized and are ionized by collision with high energy electrons. The resulting ions are extracted into a mass spec- trometer which separates them based on the ratio of the molecular mass to ionic charge. Individual ions give rise to an electronic signal which is converted via a series of internal amplification stages and external calibration standards to a mass concentration. The experimental measurements provide size resolved chemical information for a collection of particles. The AMS can detect and quantify any species that is vapor- ized at 600°C. The AMS identifies particle-bound ammonium, chloride, nitrate, sulfate, water (though only partially due to its high vapor pressure), and various organic species in ambient air samples. On some occasions, it has detected cer- tain metal species including lead, lead oxides, zinc, and zinc oxides. A wide range of organic materials have been identi- fied in ambient air (Dzepina et al. 2007) and vehicle exhaust (Canagaratna et al. 2004). Aviation PM contains appreciable amounts of organic material and sulfate. The sulfate origi- nates directly from sulfur compounds present in the jet fuel. The combustion process converts fuel sulfur compounds into SO2 quantitatively. Post-combustion, about 0.5% to 5% of the SO2 becomes oxidized to SO3 (Lukachko et al. 2008) within a fraction of a second. In the presence of water, SO3 condenses as H2SO4 on the time scale of seconds to minutes for ground level emissions. The composition of the organic material is more complex. Detailed studies (Timko, Onasch et al. 2008) have identified engine oil and partially burned hydrocarbons as the two primary types of organic materials in aviation PM. All engines studied during the APEX missions have emitted engine oil. Engine oil is a larger fraction of the total organic PM at high powers, when efficient combustion drives the emission of unburned hydrocarbons (UHCs) to near zero. Engine oil constitutes about 50% of the organic PM at high power, though for some engines the figure is closer to 90%. At low power, engine oil constitutes roughly 20% of the organic PM. Understanding that the organic content of avi- ation PM is divided between engine oil and partially burned hydrocarbons should aid future engine design efforts to reduce PM. Likewise, the potential human health and envi- ronmental impacts of partially burned hydrocarbon parti- cle contributions are likely to be different from those of engine oil. Circumstantial evidence suggests that the organic and sulfate components are internally mixed in aviation PM— in other words, each individual particle contains roughly the same fraction of sulfate and organic material as the next. The evidence for internally mixed particles comes in the form of size-resolved chemical data which show that the size distributions for organic and sulfate materials occur over 47 0 10 20 30 40 50a. b. c. D ge om (n m) -7B22 -3B1 -2C1 0 1 2 3 4 5 6 El n (1e 15 /kg _fu ) -7B22 -3B1 -2C1 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0 20 40 60 80 100 Power (%) El m (g /kg _fu ) -7B22 -3B1 -2C1 Figure D1.4. Mean particle diameter, number-based emission index, and mass-based emission index as a function of engine power for the -2C1, -3B1, and -7B22 models of the CFM56 class of engines.

48 Figure D1.5. Ranges of measured mean particle diameter for engines emissions sampled in the near field plume. Figure D1.6. Ranges of measured number-based emission index for engines emissions sampled in the near field plume.

the same size range and have similar shapes (Onasch et al. 2008). Engine operating condition (Timko, Onasch et al. 2008) and fuel sulfur content (Timko, Onasch et al. 2008; Onasch et al. 2008) influence the chemical composition of aviation PM. At idle, aviation PM seems to be predominantly organic material, a result of the relatively lower combustor efficiencies at low power conditions. At climb-out and take-off, combustor efficiency is greater than 99%, very little UHC exits the engine, and the particles contain roughly equal amounts of sulfate and organic material. As stated previously, much of the organic material emitted at climb out/take-off appears to be due to engine oil, which has nothing to do with the combustor. APEX1 provided an excellent opportunity to understand the effects of fuel sulfur content on aviation PM chemical com- position. As might be expected, increasing fuel sulfur content increases the mass of sulfate particles emitted (Onasch et al. 2008). The effect of fuel sulfur content on sulfate emissions is also seen in the JETS APEX2 and APEX3 data (Timko, Onasch et al. 2008). More surprisingly, increasing fuel sulfur content also increases the amount of organic PM emitted. The organic material contained in nucleation/growth mode particles is more sensitive to fuel sulfur content than that coated onto soot. It seems likely that condensed sulfate acts as nucleation sites for conversion of organic materials to the particle phase. D.1.9 PM Measurement Methodology Development The campaigns provided the opportunity to improve air- craft PM characterization methods, resulting in a more accu- rate PM emissions data. The APEX campaigns occurred over a 2-year period and afforded much insight into sampling methodology and diagnostic techniques for aircraft PM char- acterization. The general approach had been defined in such previous studies as Howard et al. 1996; SUCCESS (Hagen et al. 1997); SONEX/POLINAT (Hagen et al. 1999; Schlager et al. 1997); EXCAVATE (Anderson et al. 2005); and NASA QinetiQ (Whitefield et al. 2002). A detailed description of the basic methodology can be found in Schmid et al. 2004. The same methods were employed throughout the campaigns but were continuously improved as a result of lessons learned from each campaign. The lessons learned have led to subse- quent focused studies sponsored by the FAA, NASA, and DoD to refine further the methodology employed in the APEX studies. These studies are being used extensively by the SAE E-31 committee (aircraft exhaust emissions measure- ments) to develop standards for measuring PM emissions from aircraft gas turbine engines. SAE E-31 also interacts strongly with the ICAO Committee on Aviation Environ- mental Protection (CAEP) to ensure that all standards for 49 Figure D1.7. Ranges of measured mass-based emission index for engines emissions sampled in the near field plume.

aircraft emission measurements are developed for interna- tional application. D.2 Gas-Phase Data Review Summary In addition to the PM measurements described in Section 6.1, the APEX studies included a variety of gas-phase emissions measurements. The purpose of the gas-phase measurements was to complement the existing ICAO emissions databank. Whereas ICAO certification requires reporting of total UHCs and total NOX emissions, the APEX measurements provide much more in-depth chemical-level detail. Potential health and environmental implications depend critically on the prop- erties of the emitted compounds, and the APEX measurements contain more chemical information than the ICAO databank. Using the APEX data should therefore improve the complete- ness of local air quality models used for airport emissions, and may potentially improve understanding health effects. The APEX gas-phase measurements can be divided broadly into two classes: nitrogen oxides and hydrocarbons. Other gas- phase emissions are carbon dioxide, water vapor, and sulfur dioxide. Hydrocarbon emissions decrease with increasing en- gine thrust (and are most important at low power, e.g., idle/ taxi), whereas nitrogen oxide emissions are highest at high power (take-off), though are important both at low power and high power. Nitrogen oxides (NOX) are emitted prima- rily as NO and NO2. Unlike many other combustion-based engines, jet engines emit a substantial fraction of total NOX as NO2, at least at idle. A third nitrogen oxide species, HONO, is also emitted in smaller quantities (Wormhoudt et al. 2007). Aviation engines emit a wide range of hydrocarbons, the most prevalent being ethylene (C2H4) and formaldehyde (HCHO). Benzene, toluene, styrene, acetaldehyde, methanol, and naphthalene are some of the other UHCs that are emitted by aircraft engines and were measured during the APEX meas- urement campaigns. Figure D2.1 shows representative gas- phase EIs measured for a CFM56-3B1 engine during APEX3. Some features of EIs are well known: at idle, EI CO dominates, while EI NOX (defined as the sum of EI NO and EI NO2) dom- inates at take-off. The exceptional aspect of Figure D2.1 is the degree of chemical detail which far exceeds that available in ICAO certification data. Separate measurement of all hydrocarbon species emitted by gas turbines to certify commercial aviation engine models would be a daunting task. Fortunately, one of the useful find- ings of the APEX studies is that emission of the various hydro- carbon species scale with one another, regardless of engine type or thrust setting. Even when the absolute magnitudes change by an order of magnitude or more (due to changes in engine power condition, for instance), the ratio of one hydro- carbon to the next remains constant. In addition to greater chemical detail, the APEX studies provide data on the effects of two important operational vari- ables which impact aviation emissions: (1) the ambient tem- perature, and (2) emissions at power conditions other than ICAO idle (7%), approach (30%), climb-out (85%), and take- off (100%). To remove variability and provide a common baseline, the ICAO emissions databank tabulates data at stan- dard day conditions (15°C, 760 torr, 60% RH). Deviations from standard day conditions are common during normal airport operations, and the APEX data can help link actual ambient conditions to emissions performance. Likewise, air- craft engines operate at power conditions not tabulated in the ICAO databank. In fact, the runway studies conducted at Atlanta-Hartsfield airport (Herndon et al. 2008) suggest in- frequent use of ICAO power—especially 7% idle. The APEX studies measured emissions at a number of additional power conditions, including a low-power idle (4%) and several in- termediate power conditions (15%, 40%, and 65%). Use of the more complete data set should enable generation of more accurate airport emissions inventories. Moreover, the emissions effects of potential operational changes (e.g., reduced-thrust take-off, minimized idling times) can be calculated using the APEX data so that alternative scenarios can be explored before policy changes are instituted. This section begins with a description of the reproducibility of the gas-phase APEX measurements and—when possible— direct comparisons to ICAO certification data. Experimental details are provided to explain how the measurements were made. Representative gas-phase data are presented to provide a better understanding of the range of data available. Follow- ing the general discussion of the data, several potential sources of systematic error are presented. The next two sections de- scribe the hydrocarbon and nitrogen oxide in greater detail. The final section provides two examples of how the APEX data can be used: (1) effects of reduced thrust take-off and extended idle time on NOX emissions, and (2) estimating uncertainties in hydrocarbon emissions due to ambient tem- perature effects and uncertain power condition. D.2.1 Gas-Phase Data Reproducibility and Comparability to ICAO During the APEX campaigns (i.e., the combined APEX1, JETS APEX2 and APEX3 campaigns) emissions data were col- lected at over 1,200 stable conditions (i.e., the power setting/ fuel flow was stable at the desired set point) for 21 gas-phase species. Table D2.1 lists all of the gas-phase species measured during the APEX campaigns, along with the instruments used to make the measurements. State-of-the-art instruments were used for the measurements, and the quality of the instru- ments is reflected by their fast time responses and low detection limits. 50

During an emissions test, the airplanes remained grounded and chocked during all tests while the engine thrust was varied to simulate operation at ground idle (4%), idle (7%), taxi (30%), climb-out (85%), take-off (93%), and intermedi- ate power conditions including 15%, 45%, and 65% rated thrust. The power cycle during a typical experiment was as follows: (1) the engines were allowed to warm up for roughly 5 to 10 min; (2) measurements commenced as the engines were operated at ground idle; (3) the test continued as the power was increased in a step-wise fashion (e.g., 4% to 7% to 15%, etc.) up to either take-off power or climb-out power; (4) the power was directly reduced to either idle or ground idle; (5) after several minutes at idle, the power was increased di- rectly to either take-off or climb-out; (6) the test concluded as the power was reduced step-wise back to ground idle. Exhaust gas samples were taken continuously throughout the entire experiment and each stable point lasted for 2 to 5 min. Sampling was performed both at 1 m [3 ft] from the engine exit plane and further downstream of the engine (15 m, 30 m, 45 m, or 50 m [49 ft, 98 ft, 148 ft, or 164 ft] depending on the size of the engine). During each engine test, EI measurements were made at a given thrust rating both as the engine thrust was increased and as it was decreased to the set point. With a notable ex- ception, no systematic biases or hysteresis effects were found; EI CO for the RB211-535E4-B engines was 18 g kg−1 when the power was rapidly decreased from climb-out or take-off to ground idle, as compared to 35 g kg−1 when ground idle power was approached gradually in step-wise fashion. EI CO for the RB211-535E4-B did not exhibit hysteresis at power set- tings other than 4%, though the power was never rapidly 51 CO NO2 UHC NO HCHO C2H4 acetaldehyde propene 1-ring aromatics 2-ring aromatics total gas-phase EI = 42 ± 5 g kg-1 (a) idle NO NO2 CO UHC HCHO C2H4 acetaldehyde propene 1-ring aromatics 2-ring aromatics total gas-phase EI = 20 ±4 g kg-1 (b) take-off Figure D2.1. Gas-phase emissions measurements for a CFM56-3B1 (N14324, APEX3) at (a) idle and (b) take-off powers. Area corresponds to the mass-based EI. All data meaured at 30 m [98 ft]. Total gas-phase pollutant EIs provided in figure. Benzene, toluene, styrene, and phenol are the 1-ring aromatics included in this figure. Naphthalene, methylnaphthalene, and dimethylnaphthalene are the 2-ring aromatics included. The level of chemical detail provided by the APEX measurements far exceeds that available from ICAO certification data, which should lead to improved understanding of the potential impacts of airport operations on air quality.

have been corrected for this effect (Wey et al. 2006). Each EI reported in Table D2.2 is the average of all available replicate points taken at a given set of conditions. (For a given airframe/ engine combination, duplicate measurements are those made at the same downstream distance and engine power condition.) Accurate estimation of the measurement uncertainty is criti- cal to the proper use of the APEX data set. Typically, each EI reported is the average of between 3 and 6 replicate measure- ments. Experimental uncertainties were estimated in two ways. In the first method, the uncertainty was set equal to the standard deviation of all available data points at a given set of conditions. In the second, the uncertainties in each of the ex- perimental quantities which define an EI (e.g., uncertainty in the CO2 and pollutant concentrations in the ambient air and exhaust gas) were propagated as a Taylor series expansion fol- lowing a standard method. The larger of these two estimates of the uncertainty are reported here. These data handling procedures ensure that reported errors accurately represent 52 dropped from take-off or climb-out to any powers except ground idle. The RB211-535E4-B EI HCHO and some of the particulate measurements (e.g., total number count) followed similar trends as EI CO. None of the other APEX engines exhibited the engine warm-up/cool-down hysteresis effect and the extent of the phenomenon is unknown. Table D2.2 lists EI CO, EI NOX, and EI HCHO (a hydro- carbon shown to be representative for most hydrocarbon emissions) for all of the engines studied in JETS APEX2 and APEX3. The EI data presented in Table D2.2 are reproduced from a similar table originally provided by Timko, Onasch et al. (2008). Idle/taxi (either 7% or 8% depending on the engine), approach (30%), and climb-out (85%) power conditions are emphasized in Table D2.2 to facilitate comparison to ICAO certification data. EIs for take-off power conditions are only available in certain cases due to the difficulty in operating stationary aircraft at full-rated thrust. Relative humidity can in- fluence NOX emissions, and the EI NOX reported in Table D2.2 Instrument Species Detected Detection Limit Time Resolution Campaign(s) Deployed LICOR IR Gas Analyzer (models 6262 and 820) CO 2 3 ppm 1 sec APEX1,2,3 Siemens (Ultramat 23) CO CO 2 O 2 <25 ppm <500 ppm <2,500 ppm 1 sec APEX1,2,3 Multi-Gas Analyzer (MGA) (AFR-2010) N 2 O, CH 4 , HCHO, C 2 H 4 , CH 3 OH, HCOOH, Jet Fuel, SO 2 , H 2 O, CO 2 , CO, NO species dependent ranging from 0.1 (HCOOH, C 2 H 4 ) to 85 (H 2 O) 1 sec APEX1,2,3 QCL-TILDAS a NO 2 C 2 H 4 CO HCHO 0.5 ppb 2 ppb 5 ppb 1 ppb 1 sec 1 sec 1 sec 1 sec APEX2,3 APEX2,3 APEX2,3 APEX3 TILDAS b HCHO NO HONO 0.5 ppb 1 ppb 1 ppb 1 sec 1 sec 1 sec APEX1,2,3 APEX3 APEX1 Eco-Physics (CLD 700 EL ht) NO NO X >50 ppm >50 ppm APEX1,2,3 NO x Chemi- NO 0.2 ppb 1 sec/20 APEX2,3 luminescence Analyzer ThermoElectron model 42C NOYc 0.2 ppb sec 20 sec APEX2,3 Signal (300 HM) UHC >40 ppm 1 sec APEX1,2,3 PTR-MS d acetaldehyde, propene, benzene, toluene, styrene, C 2 - benzene e, phenol, naphthalene, meth ylnaphthalene, di meth ylnaphthalen e 5 ppb 8 sec APEX1,2,3 a quantum-cascade tunable infrared laser differential absorption spectrometer (Nelson et al. 1998) (Aerodyne Research, Inc.). b tunable infrared laser absorption spectrometer (Nelson et al. 2002) (Aerodyne Research, Inc.). c NOY implies NO + NO2 + RNO2 + RONO. In practice, NOY = NOX = NO + NO2 for these experiments. d proton-transfer reaction mass spectrometer (Hansel et al. 1995). e C2-benzene implies o-xylene,m-xylene, p-xylene, and ethylbenzene. Table D2.1. Gas-phase instruments deployed in APEX missions.

EI CO c (g kg -1 ) EI NO X d (g kg -1 ) EI HCHO e (mg kg -1 ) Engine Model/ Tail Number Engine Location j 00 a F 00 b (kN) 7% 30% 85% 7% 30% 85% 7% 30% 85% CFM56-2C1 f /N817NA stbd m 36±5 5±0.5 1.6±0.2 3.8±0.6 8.2±0.6 16.0±0.6 380±140 80±20 15±10 CFM56-3B1 (ICAO) 22.44 89.41 34.4 3.8 0.95 3.9 8.3 15.5 2,280 80 50 CFM56-3B1/N353SW stbd m 30±1 3.6±0.5 1.0±0.3 2.8±0.6 7.0±0.4 17±1 164±6 1.7±0.5 port 28±2 h 3.7±0.1 h 3.4±0.1 h 7.8±0.3 h 283±17 h ND g CFM56-3B1/N14324 stbd m 34±1 4±1 1.4±0.1 3.3±0.2 6.8±0.4 14±1 540±170 22±3 CFM56-3B1/ N70330 stbd m 40.0±0.8 5.1±0.3 1.6±0.1 2.99±0.07 13.2±0.7 ND g CFM56-3B1/N695SW stbd m 27±1 4.1±0.5 1.5±0.3 2.9±0.4 6.5±0.6 17±2 528±39 11.5±0.8 0.5±0.4 port 28±1 h 3.8±0.3 h 3.2±0.1 h 7.0±0.7 h 410±30 ND g CFM56-7B22 (ICAO) 24.41 100.97 22.8 2.50 0.60 4.50 10.00 19.00 2,500 100 100 CFM56-7B22/N435WN stbd m 24±7 1.9±0.7 0.40±0.03 4.3±0.3 9.5±0.6 19±4 270±40 15±8 7.3±0.5 port 23.3±3.1 h 1.71±0.05 h 4.2±0.8 h 11±1 h 380±60 h 12±7 h CFM56-7B22/N429WN stbd m 19±4 1.2±0.1 0.6±0.2 4.2±0.3 10.1±0.5 24±5 280±50 2.5±0.8 1.8±0.2 port 1.5±0.05 h 9.1±0.3 4.3±0.7 RB211-535E4-B l (ICAO) 27.9 191.7 18.24 2.43 0.26 4.58 8.65 19.3 140 5 0 RB211-535E4-B l /N75853 stbd m 18±8 2.9±1.4 0.22±0.08 5±1 9.3±0.8 24±5 80±2 ND g 11±3 RB211-535E4-B l /N74856 stbd m 19±1 2.1±0.2 0.20±0.03 5.0±0.6 10±1 23.9±0.7 219±9 ND g PW4158 (ICAO) 30.7 258.0 20.99 1.88 0.54 4.8 11.8 23.7 1,780 140 2 PW4158 k / N729FD stbd m 39±3 2.2±0.5 3.5±0.3 9.6±0.8 22.4±2 1,010±70 ND g AE3007-A1E (ICAO) 19.06 37.16 37.97 5.63 0.64 4.26 7.42 14.91 AE3007-A1E i /N11193 stbd m 29±12 3±1 0.267±0.004 3.7±0.7 7.7±0.5 13.2±0.6 400±100 12±1 port 35±1 4.4±0.1 0.30±0.01 3.43±0.09 7.3±0.2 12.1±0.7 660±20 ND g ND g ND g ND g ND g AE3007-A (ICAO) 18.08 33.73 33.73 17.35 3.28 3.83 7.79 17.47 AE3007-A i /N16927 stbd m 32.8±0.7 4.0±0.4 0.33±0.03 3.4±0.3 6.7±0.5 10.4±0.8 520±20 ND g 17.1±1.2 CJ6108A i,j /N616NA stbd m 140±7 45±7 21±2 2.4±1.6 3.2±0.5 4.6±0.3 2,500±500 400±130 47±11 a 00 = pressure ratio. b F00 = rated thrust (kN). c EI-CO measurements averaged over all available downstream sampling locations. d EI-NOX measurements averaged over all available sampling locations. EI-NOX equals the sum of EI-NO2 and EI-NO, both in units of mass of NO2. e EI HCHO reflects average value of all available 1-m data. ICAO values are for UHCs—EI-UHC. ICAO does not speciate hydrocarbon data. f The CFM56-2C1 engine was mounted on a NASA airframe not used in the commercial fleet. No ICAO data are available for this engine. g ND indicates that the species concentration was less than the detection limits of the instrument. Blank spaces indicate missing data. h Measurements made at 1-m sampling location used for this data point— no downstream data available. i The 7% thrust setting was not studied for this engine—data measured at 8% rated thrust. j The CJ6108A thrust is lower than the ICAO certification limit. No ICAO data are available for this engine. k The 85% thrust condition was not studied for this engine—data measured at 80% thrust. l The RB211-535E4-B engines studied during the APEX missions were equipped with the Phase V combustor. mThe abbreviation “stbd” refers to the “starboard” engine. Likewise, “port” is the port engine. Data excerpted from Timko, Herndon et al. 2008. Table D2.2. Summary of EIs measured for some gas-phase species during APEX.

experimental reproducibility. Sources of systematic error are considered later in this section. A number of internal and external consistency checks were used to help build confidence in the gas-phase data. Specifi- cally, when multiple instruments were used to measure the same quantity (i.e., NOX), the various measurements typically agreed to within the specified limits of instruments uncer- tainty. Moreover, data measured for multiple examples of the same engine technology agree within the limits of experimen- tal reproducibility. When such comparisons are possible, the APEX data are in good general agreement with ICAO certifi- cation data. These two comparisons establish the quality of the APEX data. More details on the two comparisons are pro- vided in the following two paragraphs. Qualitatively, the EIs behave as expected. As engine power increases, so too do combustor efficiency and EI NOx, whereas EI CO and EI HCHO decrease. Quantitatively, EI CO and EI NOX data for the 4 CFM56-3B1 engines agree nearly to the limits of experimental uncertainty. EI CO for the CFM56- 3B1/N70330 engine is an apparent outlier. Similarly, duplicate measurements of EI CO and EI NOX for the other engines agree reasonably well when such data are available. EI HCHO measured for different examples of a given engine type, how- ever, is much more variable than EI CO or EI NOX. For ex- ample, EI HCHO takes values ranging from 164±6 (N353SW, starboard) to 540±170 mg kg−1 (N14324, starboard) for CFM56-3B1 engines at 7% rated thrust. Ambient tempera- ture is the likely source of much of the EI HCHO variability, and this topic is discussed in the section on hydrocarbon emissions. At high powers, the concentration of HCHO was frequently lower than the detection limits of the instrument (about 0.5 ppb). The abbreviation “ND,” short for “not detected,” indicates that the concentration was below the de- tection limits. (Table D2.1 lists detection limits of the various instruments.) EI NOX is generally highest for the larger engines (i.e., PW4158 and RB211-535E4-B), which is expected as the certification process makes allowances for maximum rated thrust. EICO does not follow the engine size trend and ranges from about 20 g kg−1 (RB211-535E4-B) to about 30 to 40 g kg−1 (CFM56) for the turbofan engines at 7% thrust. The single turbojet engine that was tested (CJ6108A/N616NA) has the lowest EI NOX and the highest EI CO, by a factor of about 2. It has the highest EI HCHO by about a factor of 5. The low fuel efficiency and inlet combustor temperature of the turbojet CJ6108A accounts for the high EI CO and EI HCHO and low EI NOX. The good agreement for EIs measured for different examples of the same engine technology (e.g., CFM56-3B1) indicates that the quality of the data is good and can be trusted for calculations. The CJ6108A is an older engine technology; newer turbojets may perform differently. Comparison of measured EIs to ICAO certification data helps understand the quality of the APEX data. ICAO EIs for the APEX engines are provided in Table D2.2 for comparison with the data collected during the measurement campaigns. The APEX data set generally agrees with ICAO EIs. The good overall comparison between APEX data and ICAO certifica- tion data further underlines that the experimental data are high quality. Some important exceptions and qualifiers are required, as follows: • Field measurements of EI CO agree with the ICAO values within the limits of experimental uncertainty for most of the engines, though the variability in the data is large enough (generally about 25%) to obscure small differences. • Discrepancies between APEX EI CO and ICAO EIs generally occurred only when the actual fuel flow rate deviated from the ICAO value. Measurements of EI CO for the AE3007 engines were higher than ICAO EI CO, but the experimen- tal fuel flow rate was lower than the ICAO value. • Measured EI NOX is generally lower than the ICAO value by about 10%. Given known experimental uncertainties and biases, 10% can be considered to be good agreement. • EI NOX increases monotonically for all engines in the APEX data set. • Discrepancies between APEX EI NOX and ICAO EIs gen- erally occurred only when the actual fuel flow rate deviated from the ICAO value. D.2.2 Gas-Phase Measurement Reliability and Sources of Systematic Error In addition to reproducibility, absolute measurement accuracy is also important. Instrument detection limits are typically on the order of 1 ppb to 10 ppb. Typical measured concentrations are on the order of 10 ppb to 1,000 ppb. Therefore, the uncertainties of EIs less than about 0.05 g kg−1 are greater than experimental variability. The uncertainty is imposed by detection limits, such as hydrocarbons (especially at high thrust conditions) and HONO. Additionally, we have identified several other sources of systematic errors, as follows: 1. EI CO is lower when measured at the engine exit plane than when measured “downstream” by much as 20% at “ground” idle conditions (4% power) and by as much as 10% at “ICAO” idle (7%). Two likely mechanisms have been identified: (1) errors in the fitting procedure used to convert absorbance signals into concentrations for highly concentrated samples (>5,000 ppb). The fitting error is estimated to be no more than 5%, which is not sufficient to explain the observed discrepancies by itself; (2) oxidation of CO into CO2 in the 1-m probe used to sample engine exit plane gases. NO2, OH, O, and HO2 are likely CO oxi- dants either in the gas-phase or on the heated metal surfaces in the 1-m probe. The apparent loss of CO at the engine exit 54

plane is most notable in the APEX3 data, consistent with the fact that the 1-m sample rake was not cooled during this mission, unlike the others. Because of the uncertainties in the engine exit plane CO measurements, only downstream CO measurements are included in Table D2.2. 2. Certain organic acids, namely formic and acetic acids, have been observed at downstream sample rakes. The presence of organic acids is due to chemical reactions which occur in the downstream probe and are not indicative of actual en- gine emissions. EIs for organic acids are not reported here. 3. NO2 may be converted to NO in the 1-m probe, especially at idle conditions. The NO2/NO transformation in the 1-m probe may be chemically coupled to the CO/CO2 con- version mentioned previously. Thus, NO2 may indirectly serve as the oxidant for conversion of CO into CO2. Fur- thermore, evidence suggests that NO may be converted into NO2 in the exhaust plume with a reaction time scale on the order of seconds (Wood et al. 2008). The two con- version effects may partially balance one another making quantitative speciation of NOX into NO and NO2 challeng- ing when measured at the engine exit plane. Just as with EI CO, the deviation between engine-exit-plane and down- stream EI NO2 data is most pronounced for measurements made with the uncooled 1-m probe used during APEX3. Also, HONO may either be formed or destroyed in the 1-m sampling probe and this chemistry is poorly understood. 4. Butanol is used for certain particle size measurements. During transitions from one power to the next, butanol can be ingested into the internal transfer lines and carried from the particle instruments to the gaseous instruments. Butanol in the internal transfer lines gives rise to a false signal in the hydrocarbon measurement instrumentation (specifically a m/z = 57 signal in the PTR-MS) that prevents quantification of the important pollutants, butadiene and acrolein. Efforts are underway (Knighton et al. 2007a) to discriminate between butanol, butadiene, and acrolein. D.2.3 NO2, NO, and Total NOX Emissions Most combustion sources emit NOX primarily in the form of NO (Heywood 1988). Gas turbine engines operating at low power conditions are an exception, as they emit a substantial amount of NOX as NO2. Figure D2.2 plots the nitrogen oxide EIs for a RB211-535E4-B (a turbofan) and a CJ6108A (a turbo- jet). Figure D2.2a is representative of all of the turbofan engines considered in the APEX missions. Figure D2.2b is shown for contrast and the data may be relevant for general aviation engines. Some general features are apparent in the NOX data: • For the turbofan, NO2 is a significant fraction of the total NOX at low thrust (idle) conditions. In fact, 99% of the 55 NO NO 2 NO NO 2 (b) 30 25 20 15 10 5 0 EI (g kg - 1 ) 100 80 60 40 20 0 Power % 30 25 20 15 10 5 0 EI (g kg - 1 ) 10080 60 40 20 0 Power % CJ6108A (N=1) (a) RB211-535E4-B (N=2) Figure D2.2. NOx speciation for (a) RB211-535E4-B and (b) CJ6108A engines. ICAO data points () are shown for reference, when available. The CJ6108A thrust is below the ICAO threshold for certification, and no ICAO data are available for this engine. Both EI NO and EI NO2 are plotted in units of NO2 mass equivalents for direct comparison with ICAO EI-NOX. The data shown in this figure are representative of the entire data set. The experimental data agree with ICAO EI NOX for all engines except the AE3007s at high thrust. The data are not corrected for ambient humidity, though this effect was found to be small (<6%). N denotes the number of engines studied.

NOX emitted from certain CFM56 engines was in the form of NO2 at ground idle (4% rated thrust). • EI NO2 steadily decreases and EI NO steadily increases as power is increased for the turbofan engines. Less than 10% of the total NOX is emitted as NO2 at high power condi- tions for the turbofan engines. • The turbojet engine, which is shown for contrast, emits a substantial fraction of the NOX as NO2 at all power conditions. Wormhoudt et al. (2007) provide more details on APEX1 NOX data, including HONO data. Wood et al. (2008) de- scribe APEX2 and APEX3 NOX data, both as measured dur- ing dedicated engine tests and from advected plume studies. D.2.4 Speciated Hydrocarbon Emissions The APEX data set includes EIs for a number of hydrocar- bons including formaldehyde, ethylene, acetaldehyde, ben- zene, and styrene (see Table D2.1 for a complete list). The major conclusions that can be drawn from the hydrocarbon data are described below: • Hydrocarbon EIs are highest at low thrust conditions and each individual EI falls to values below 0.1 g kg−1 at thrusts above 15%. • Qualitatively, EI HCHO follows the same trends with re- spect to engine thrust as does EI CO. • Formaldehyde and ethylene are the most prevalent hydro- carbons emitted from gas turbine engines on an EI basis. • The sum of all measured hydrocarbon EIs is within about 10% to 20% of the value of UHC EIs measured by the ICAO flame-ionization detection method (FID). The quan- titative agreement between the speciated measurements and the FID indicate that many of the UHCs have been in- cluded in the speciated analysis, yet the direct comparison is difficult since the FID is not equally sensitive to all HCs (aldehydes are under quantified) and not all HC species were separately quantified. • Hydrocarbon EIs decrease by a factor of 100 or more as the power condition is adjusted from idle to take-off. Over almost the entire range, hydrocarbon EIs vary in proportion to one another. Therefore, accurate measurement of one hydrocarbon EI may allow quantification of all other hy- drocarbon EI, provided a consistent speciation profile. The mutual scaling may not hold when the EIs fall below 0.1 g kg−1, though this distinction has little bearing on emissions inventories. • Ambient temperature strongly influences hydrocarbon EIs. A 20°C decrease in ambient temperature resulted in a 10-fold HCHO increase (when power setting is used as the scaling variable) or a 3-fold HCHO increase (when fuel flow rate is used as the scaling variable). The final two points are especially important and are de- scribed in more detail in the following sections. Mutual scaling of the hydrocarbon EIs. One of the major findings of the APEX series of experiments is the mutual scaling of the various hydrocarbon EIs. As power conditions and/or ambient conditions change, all of the hydrocarbon EIs change in concert so that the ratios of the hydrocarbon EIs with respect to one another are constant. Typically, HCHO is used as a reference, a selection originally made because HCHO is measured by a separate instrument than the majority of the hydrocarbons (see Table D2.1). Fig- ure D2.3 contains plots of APEX1 measurements of EI HCHO and EI benzene as functions of percent rated thrust for several engine conditions and a correlation plot of EI benzene versus EI formaldehyde for all available engine con- ditions (Knighton et al. 2007b). Spicer et al. (1994) performed measurements of hydrocarbon EIs and their data is shown for comparison. The highly correlated benzene-formaldehyde plot shows that, even though the individual measurements have substantial variability (e.g., due to changes in ambient temperature), the ratio of benzene to formaldehyde remains constant. The relationships depicted in Figure D2.3 suggest that measurement of one hydrocarbon EI might be used to de- termine the EIs of other unmeasured hydrocarbons. The scaling law depicted in Figure D2.3 applies to all hydrocarbons which are measured during a standard APEX experiment (see Table D2.1). Due to the sensitivities of the hydrocarbon instruments deployed during the APEX missions, alkane EIs have not yet been measured in gas turbine exhaust and we cannot confirm if alkane EIs obey the “universal” scaling law. Measurements made by Spicer et al. (1994) suggest that the overall contribution of alkanes to the total hydrocarbon EI is less than roughly 10%. Likewise, an analysis performed by Yelvington et al. (2007) suggests that the alkane contri- bution to the total UHC EI is minor. In that treatment, the individual gas-phase EIs were summed on a per carbon atom basis to yield a total hydrocarbon EI which was within about 10% to 20% of that measured using the standard ICAO FID method (Yelvington et al. 2007). Therefore, the alkane contribution to the total hydrocarbon EI is likely to be less than about 10%; however, precise numbers are not available because the FID UHC number may not represent a total HC measurement due to its non-uniform sensitivity. The effect of ambient temperature on hydrocarbon EIs. Ambient temperature strongly influences hydrocarbon EIs (but much less so NO, NO2, or CO). During APEX1, a 20°C decrease in ambient temperature (from 35°C to 15°C) resulted in a 3-fold increase in EI HCHO. (The same change in ambient temperature results in a 10-fold increase in EI HCHO when 56

power setting, rather than fuel flow rate, is held constant. In other words, lower ambient temperatures require lower fuel flow rates to achieve a desired power setting. EI HCHO increases with decreasing fuel flow rate. The relationship be- tween fuel flow rate and EI HCHO accounts for about one- third of the observed dependence of EI HCHO on ambient temperature.) Figure D2.4 is a plot of EI HCHO measured during APEX1. By virtue of the hydrocarbon scaling law, Fig- ure D2.4 is representative of all hydrocarbon emissions. The variability in the EI versus thrust plots depicted in Figure D2.3 is due to the temperature sensitivity. APEX1 was unique among the APEX series of missions as a single test engine that was studied over a wide range of ambient temperature. Quantify- ing the effect of ambient temperature in the APEX data set is more challenging, though some data are consistent with the APEX1 results. For instance, the effect of ambient temperature may be reflected in the EI HCHO comparison between N14324/CFM56-3B1 (540 mg kg−1 at 7%, 8°C) and N353SW/ CFM56-3B1 (160 mg kg−1 to 280 mg kg−1 at 7%, 13°C). Like- wise, the two RB211-535E4-B engines studied during APEX3 may show a temperature effect: EI HCHO for N75853 equals 80 mg kg−1 (17°C) while that of N74856 (10°C) is 219 mg kg−1. Yelvington et al. (2007) show that temperature variability of fuel flow rate accounts for about 1/3 of the observed variability in hydrocarbon EIs and suggest that relative humidity effects and/or instrument/sampling variability account for the rest. The likely emissions ramifications of the ambient temperature effect are clear: failure to account for the ambient temperature 57 Figure D2.3. Formaldehyde and benzene EI data measured during APEX1 (CFM56-2C1). (a) EI HCHO as a function of power condition; (b) EI benzene as a function of power condition; (c) EI benzene as a function of EI HCHO. The individual hydrocarbon EIs vary by about a factor of 10 at a given power condition (either 4% or 7%), and by more than that between the two power conditions shown here. The apparent variability in individual EIs is captured as a strong correlation in the scatter plot. All hydrocarbons measured thus far vary in proportion to one another—benzene/HCHO provides a representative example. Comparison data provided by Spicer et al. (1994).

effect may lead to estimated EIs which are inaccurate by a factor of 10 or more. The effect of ambient temperature is clear—decreasing ambient temperature by 20°C results in a 10-fold increase in EI-HCHO. Data for power conditions >30% are omitted as the EIs are small (<0.01 g kg−1) and noise in the measurment sometimes exceeds the absolute value. D.2.5 Potential Use of APEX Data and the ICAO LTO Cycle to Generate Emissions Inventories The APEX data can be used in conjunction with airport op- erations data to generate airportwide emissions inventories. The depth of chemical information and the wider range of op- erational conditions included in the APEX data set allow it be used to generate more comprehensive emissions inventories than is possible with ICAO data. Wood et al. (2008) demon- strate the use of APEX data for estimating emissions during LTO cycles, and the Wood approach is adopted here. Table D2.3 contains the results of a sample calculation for the nitro- gen oxide emissions of a CFM56-3B1 engine calculated over the standard ICAO LTO cycle. Several observations can be made: • APEX data and ICAO data yield similar estimates for total NOX emissions; • Almost 20% of the total NOX is emitted as NO2; • Most of the NO2 is emitted during idle; • About half of the total NOX is emitted during climb-out. In addition to applying the APEX data set to standard LTO cycles, the emissions of various pollutants in hypothetical sce- narios can be calculated. In Table D2.4, the total NOX, NO2, CO, and HCHO emitted during several hypothetical LTO cycles are listed for a CFM56-3B1 engine. The first two rows of Table D2.4 present data for the standard ICAO LTO cycle, using either ICAO or APEX EIs. The final NOX/NO2 figures presented in Table D2.3 can be compared directly to the first two rows of Table D2.4. The difference in APEX and ICAO estimates of CO emissions is due to a discrepancy in the EIs at 7% (28.1 g/kg for APEX compared to 34.4 g/kg for ICAO). Each row subsequent to the second lists emission estimates with one LTO parameter changed from the default. • Row 3: Reduced power idle. HCHO emissions increase by about 40% and CO by nearly 20%. • Row 4: Prolonged idle. Doubling the idle time to 52 min increases NOX emissions by only 20% but increases NO2 emissions by 67%. HCHO and CO emissions roughly double. APEX data also capture variability in emissions data. Depending on the application and on the analysis technique used to interpret the data, real emissions variability data can be very useful for understanding the range of emissions that can realistically be expected during the course of normal operations. Primary sources of variability may include experimental errors, engine age and maintenance history, and ambient conditions. 58 0.01 2 4 6 8 0.1 2 4 6 8 1 2 EI H CH O (g kg - 1 ) 0.1 2 3 Fuel Flow Rate (kg sec-1) 4% 5.5% 7% 15% 30% 35 30 25 20 ambient temperature (°C) APEX -1 CFM56-2C1 Figure D2.4. Formaldehyde emission index (EI) as a function of fuel flow rate measured during APEX1 for a CFM56-2C1 engine. Power condition is indicated directly on the graph. LTO Phase Time in Mode (min) Fuel Flow Rate (kg/s) NOX EI (g/kg) NO2 EI (g/kg) Total NOX (kg) Total NO2 (kg) Approach 4 0.306 6.9 0.98 0.51 0.072 Idle 26 0.114 2.98 2.98 0.53 0.39 Take-off 0.7 0.953 16.8 1.2 0.67 0.048 Climb-out 2.2 0.886 15.1 1.15 1.77 0.13 Totals APEX 3.5 0.64 ICAO 3.6 n/a Table D2.3. Emission indices and engine parameters used to calculate the total NO2 and NOx emissions from a CFM56-3B1 engine during a standard landing take-off cycle.

For the current stage of measurement development and for the number of engines tested, discerning one potential source of variability from the next is not always possible. As APEX style measurements become more routine, experimental variability will be reduced so that engine-to-engine variability can be iso- lated. In the meantime, the effects of ambient conditions, especially ambient temperature, and power conditions are clear. Figure D2.5 captures data variability graphically for total UHC emissions. In Figure D2.5, the UHC emission rate (i.e., mass of hydrocarbons emitted per second) is plotted over the course of a standard LTO (where the idle time has been re- duced from 26 min to 22 min). The area under the curve is pro- portional to the total quantity of emitted UHC. The variability in the UHC emissions has been calculated based on variability in ambient temperature (here, a 20°C range has been consid- ered) and to account for reduced power idle (4% compared to 7%). Readily apparent is that aircraft engines emit most— almost 90%—of the hydrocarbons during idle, a consequence of both the EI and the time in mode. The APU hydrocarbon emissions are shown for comparison and are clearly negligible compared to those of the aircraft engines. Also apparent is the high degree of variability for the estimated idling emissions. The errors bars in the diagram represent an estimate of the uncertainty in the true UHC emissions rate which is greatly impacted by ambient temperature and the actual thrusts used. Detailed analyses which take into account this real emissions variability will provide more realistic emissions inventories for use in chemical dispersion models. 59 Figure D2.5. Total UHC emission rates plotted as a function of time in mode during LTO cycle. The area under the curve (i.e., the area of the “boxes”) is proportional to the total amount of UHC emitted during a portion of the cycle. About 90% of the UHCs are emitted during idling. The error bars are an estimate of the range of the emission rate and account for uncertainties in the true thrust values used (e.g., 4% idle vs. ICAO 7%) and uncertainties in the influence of ambient temperature on hydrocarbon EIs (see Figure D2.4). Row # Scenario Total NOx(kg) Total NO2 (kg) Total CO (kg) Total HCHO (kg) 1 ICAO base case (ICAO times, thrust, and EIs) 3.6 n/a 6.5 n/a 2 Base case (ICAO times and thrust levels, APEX EIs) 3.5 0.64 5.5 0.082 3 4% idle, 26 min 3.6 0.59 6.4 0.13 4 7% idle, 52 min idle time 4.3 1.1 10.6 0.16 Table D2.4. CFM56-3B1 NOX, CO, and HCHO emissions for different operating scenarios.