ACRP Web-Only Document 41: Alternative Jet Fuels Emissions: Quantification Methods Creation and Validation Report (2019)

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.

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 12 4.0 Pollutant-Specific Analysis In the following sections the detailed analysis undertaken to quantify the impacts for the pollutant species SOx, nvPM N, nvPM Mass, NOx, CO, UHC, and HAPs is presented. 4.1 SOx The SOx emission indices (EISOx, blend) of the conventional/SAJF blends are found to be directly proportional to the amount of sulfur found in the fuel burned, independent of engine type and operating condition, according to the following functional relationship: EISox,blend = %% ∗ EISox,conv (1) where EISOx, conv denotes the SOx emission index for the conventional fuel. This relationship assumes that the alternate fuel contains negligible sulfur compared to that of the conventional fuel. This assumption is borne out by the literature. Hence the impact on SOx of the SAJF can be defined as ΔSOx = EISOx,blend − EISOx,conv = − %% ∗ EISOx,conv (2) STEP 1 – Critical metrics are found to be blend% and fuel sulfur content. Based on available data from the State of the Industry Review, Δ SOx is found to be independent of engine type, engine operating condition, and atmospheric conditions. The only dependency on the alternate fuels was its sulfur content with metrics sulfur concentration ([S] in %wt. sulfur, or ppm) and blend%. STEP 2 – The SOx impacts spreadsheet is given in Table 3. Table 3: The SOx emissions impact spreadsheet Alt fuel Ref fuel Engine Impact Ref # FT GTL JP-8 CFM56-2C1 EI_SO2 ↓ 90% for pure FT, and ↓ intermediately for blends. 6 HEFA/FT JP-8 F117-PW-100 SO2 ↓ 50% for 50% blend. 20 STEP 3 – The emissions data indicated a quantifiable presence of sulfur, not only in the conventional fuel, but also in the blended SAJFs. Fuel analyses where available supported this observation. Step 3 concludes that any impact analysis should account for the total sulfur content in the blended fuel. STEP 4 – The following function was found to represent the impact of sulfur on SOx emissions. EISOx,blend = %% ∗ EISOx,conv + 𝑏𝑙𝑒𝑛𝑑% ∗ EISOx, SAJF (3)

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 13 where EISOx, SAJF denotes the SOx emission index for the neat SAJF, and ΔSOx = EISOx,blend - EISOx,conv = 𝑏𝑙𝑒𝑛𝑑%100% ∗ (EISOx,SAJF − EISOx,conv) (4) ΔfSOx,EI = ΔSOxEISOx,conv = 𝑏𝑙𝑒𝑛𝑑%100% ∗ EISOx,SAJFEISOx,conv - 1 , (5) where ΔfSOx, EI denotes the fractional impact factor for SOx expressed in terms of emission indices. Using EISOx,SAJF = EISOx,conv ∗ [SSAJF][Sconv], (6) where EISOx, SAJF denotes the emissions factor for the neat SAJF. Then ΔSOx = 𝑏𝑙𝑒𝑛𝑑%100% ∗ [SSAJF][Sconv] - 1 ∗ EISOx,conv (7) where [SSAJF] and [Sconv] denote the sulfur weight percents for the component fuels, and thus the SOx fractional impact factor ΔfSOx, [S] can be written in terms of sulfur weight percents as ΔfSOx,[S] = 𝑏𝑙𝑒𝑛𝑑%100% ∗ [SSAJF][Sconv] − 1 . (8) There will be measurement uncertainties in blend%, EISOx, and [S], which will lead to uncertainties in ΔfSOx, EI and ΔfSOx, [S], as given by the expressions: δΔfSOx,EI = ΔfSOx − 𝑏𝑙𝑒𝑛𝑑%100% 1 + 𝛿 𝑏𝑙𝑒𝑛𝑑%100% ∗ EISOx,sajf-δEISOxEISOx,conv+δEISOx – 1 , (9) and δΔfSOx,[S] = 𝛥fSOx − 𝑏𝑙𝑒𝑛𝑑%100% ∗ 1 + 𝛿 𝑏𝑙𝑒𝑛𝑑%100% ∗ [SSAJF]-δ[S][Sconv]+δ[S] – 1 . (10) where δEISOx and δ[S] denote uncertainties in measure SOx emission index and fuel sulfur content. Here the uncertainties are taken to be the difference between the impact factor calculated using the parameters EISOx,SAJF, and EISOx,conv shifted by δEISOx in such a way as to give the largest deviation, for the case of Eq. (5), and calculated using the parameters [SSAJF] and [Sconv] shifted by δ[S] in such a way as to give the largest deviation, for the case of Eq. (8).

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 14 STEP 5 – In Reference 6 (Beyersdorf, et al.), a paper describing the results of the Alternative Aviation Fuel Experiment (AAFEX) campaign, the SAJF of interest was Fisher-Tropsch (FT) Gas To Liquid (GTL) and the conventional fuel was JP-8 in a 50% blend. The fuel sulfur content of the JP-8 was reported to be 1148 ppm, and the sulfur content of the FT GTL was assumed to be zero. However, SO2 was observed in the exhaust of neat FT GTL. A companion report for project AAFEX (ref 22) provided the fuel analyses for all fuels burned in the project and revealed a fuel sulfur content of 19 ppm for the neat FT GTL. The blend% of 50% used in AAFEX applied to Eq. (5) and Eq. (9) thus gives a ΔfSOx, [S] of -0.4917 ± 0.0069, assuming a 0.5% uncertainty in blend% and uncertainties in fuel sulfur content of 10%. Using the Beyersdorf observational EI data (EISOx, conv = 2.3 g/kg; EISOx, SAJF = 0.2 g/kg) in Eq. (5) and (9) gives ΔfSOx, EI = -0.492 ± 0.036, assuming a 0.5% uncertainty in blend% and uncertainties in EI measurements of 0.15 g/kg. These measurement uncertainties are conservative estimates of uncertainty based on measurement experience. To determine if the difference in impact factors between the EI method and the Sulfur content method is significant, the Error Ratio (ER) can be defined as: 𝐸𝑅 = | , ,[ ]|, ,[ ] (11) where ER < 1 implies the error bars in ΔfSOx overlap. For the Beyersdorf data, ER=0.83. Hence the impact factor relationship, Eq. (8) and (10), agrees with the measurement data to within experimental error. In Reference 20 (Corporan, et al.), a paper describing emissions from blends of Hydroprocessed Renewable Jet (HRJ) and FT SAJFs with JP-8, burned in F117-PW-100 engines on a C-17 aircraft. A 50% blend of HRJ with JP-8 was studied. The fuel sulfur content of the JP-8 was reported to be 0.08 wt.%, and 0.02 wt.% for the HRJ. Applying Eq. (8) and (1) gives a ΔfSOx[S] of -0.375 ± 0.072, assuming a 0.5% uncertainty in blend% and uncertainties in fuel sulfur content of 1%. Using the Corporan normalized observational EI data (EIconv = 1; EISAJF = 0.55) in Eq. (5) and (9) gives ΔfSOx, EI = -0.225 ± 0.181, assuming a 0.5% uncertainty in blend% and uncertainties in EI measurements of 0.3. For the Corporan data, ER=0.59. Hence the impact factor relationship, Eq. (8) and (10), agrees with the measurement data to within experimental error. SOx Findings The fact that the uncertainty in ΔfSOx, [S] is smaller than the absolute value of ΔfSOx, [S] implies that there is a statistically meaningful SOx impact associated with alternative fuel usage. 𝛥𝑓𝑆𝑂𝑥, [𝑆] = 𝑏𝑙𝑒𝑛𝑑%100% ∗ [SSAJF][Sconv] − 1 . 𝛿𝛥𝑓𝑆𝑂𝑥, [𝑆] = 𝛥𝑓𝑆𝑂𝑥 − 𝑏𝑙𝑒𝑛𝑑%100% ∗ 1 + 𝛿 𝑏𝑙𝑒𝑛𝑑%100%∗ [S]sajf-δ[S][S]conv+δ[S] – 1

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 15 Illustrative example: To illustrate the use of the impact factor, assume an airport has normal SOx emissions of 1,000 kg/year. Assuming 12% of the jet fuel used at the airport is blended conventional/SAJF at a 50% blend (50% blend  Δf_ SOx = -0.375 and δΔf SOx,[S] = 0.072), and assuming [S]sajf = 0.02 and [S]conv=0.08, the SOx emissions savings would be 45 kg/year (1000*0.12*0.375) with an uncertainty of 8.6 kg/year (1000*0.12*0.072). 4.2 nvPM Number For nvPM number we anticipated a similar relationship as that for SOx. At currently approved blend percentages of up to 50%, the reduction in nvPM number emissions is directly related to the amount of SAJF burned, thus the emission index of the blend (EInPM, blend), for a given engine and operating condition, can be expressed as follows: EInPM,blend = 𝛼 ∗ EInPM,conv ∗ (1 − 𝑏𝑙𝑒𝑛𝑑%) + 𝛽 ∗ EInPM,SAJF ∗ 𝑏𝑙𝑒𝑛𝑑%, (12) where EInPM, conv and EInPM, SAJF denote the emission indices for the conventional and alternate fuels, respectively, and α and β are constants. The impact on nvPM number due to the SAJF for specific engines and their operating conditions is Δf_EIn = EInPM,blend - EInPM,conv EInPM,conv = 𝛼 ∗ 1 − 𝑏𝑙𝑒𝑛𝑑%100% + 𝛽 ∗ EInPM,ASJFEInPM,conv ∗ 𝑏𝑙𝑒𝑛𝑑%100% – 1 = 𝜑0 + 𝜑1 ∗ 𝑏𝑙𝑒𝑛𝑑%100% 𝑤𝑖𝑡ℎ 𝜑0 = 𝛼 − 1, 𝜑1 = 𝛽 ∗ EInPM,ASJFEInPM,conv − 𝛼 . (13) STEP 1 – Based on the limited available data from the State of the Industry Review, the observational data is found to be dependent on power and blend%. For use in the Aviation Environmental Design Tool (AEDT) model, an average value for impact factor, weighted over LTO cycle fuel burns, is calculated, resulting in impact factors depending on blend% alone. STEP 2 AND 3 – The nvPM N emissions impact spreadsheet is given in Table 4 below. In the spreadsheet Δf_EIn {LTO EIn} denotes the power dependent nvPM Number impact factors; FF*t denotes the product of fuel flow rate for a given operational mode and the time in mode as defined by the ICAO LTO cycle. This represents the fuel burned in the mode and is used as a weighting function to get an LTO weighted impact factor, Δf_EInweight. Δf_EInavg, weight is the average of all Δf_EInweight values recorded for a given blend%. Multiple values were only found for 50% and 100% blend percentages; δ denotes the standard deviation in Δf_EInweight values and is used as the uncertainty in Δf_EInavg, weight. STEP 4 – Development of functional impact relationships. Table 4 gives the Δf_EInavg, weight for two values of blend%, with associated uncertainty (δ). The published references gave EI values at various measured engine power points. Linear interpolations or extrapolations were performed on these to get EI values at

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 16 the LTO cycle power points. Line loss was accommodated in this analysis through its implicit inclusion in the published data. Table 4: nvPM number emissions impact spreadsheet Sp ec ie s En gi ne Co nv fu el SA JF bl en d% Re f # LT O pw r LT O EI n b le nd (a rb itr ar y un its ) LT O EI n c on v (a rb itr ar y un its ) LT O Δf _E In FF *t W t f ct Δf _E In w ei gh t Δf _E In av g, w ei gh t δ nvPM N CFM56 -2 JP-8 FT GTL 50 6 7 2.32E+14 1.17E+15 -0.80 3.128 -0.66 -0.48 0.10 30 1.49E+14 9.89E+14 -0.85 1.48 85 7.13E+14 1.45E+15 -0.51 2.488 100 9.04E+14 1.37E+15 -0.34 0.961 nvPM N GE CF700- 2-D-2 Jet A1 HEFA- SPK 50 15 7 2.90E+15 7.30E+15 -0.60 3.128 -0.54 30 3.09E+15 7.77E+15 -0.60 1.48 85 4.37E+15 9.07E+15 -0.52 2.488 100 6.97E+15 9.87E+15 -0.29 0.961 nvPM N CFM56 -7B Jet A1 FT GTL 50 36 7 2.39E+01 1.00E+02 -0.76 3.128 -0.51 30 4.37E+01 1.00E+02 -0.56 1.48 85 7.18E+01 1.00E+02 -0.28 2.488 100 7.75E+01 1.00E+02 -0.23 0.961 nvPM N PW308 JP-8 FT GTL 50 22 7 66.66421 100 -0.33 3.128 -0.26 30 7.16E+01 1.00E+02 -0.28 1.48 85 81.35593 100 -0.19 2.488 100 81.35593 100 -0.19 0.961 nvPM N TF33 JP-8 FT GTL 50 22 7 56.96892 100 -0.43 3.128 -0.33 30 62.43468 100 -0.38 1.48 85 76.19070 100 -0.24 2.488 100 79.7 100.0 -0.20 0.961 nvPM N CFM56 -2 JP-8 HEFA T 50 22 7 19.02036 100 -0.81 3.128 -0.46 30 37.76184 100 -0.62 1.48 85 84.7 100.0 -0.15 2.488 100 116.1782 100 0.16 0.961 nvPM N CFM56 -7 JP-8 FT GTL 50 22 7 18.25047 100 -0.82 3.128 -0.45 30 49.2 100 -0.51 1.48 85 87.97677 100.0 -0.12 2.488 100 98.3 100 -0.02 0.961 nvPM N T63 JP-8 FT GTLShel 50 22 7 41.09336 100 -0.59 3.128 -0.46 30 47.47716 100 -0.53 1.48

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 17 Sp ec ie s En gi ne Co nv fu el SA JF bl en d% Re f # LT O pw r LT O EI n b le nd (a rb itr ar y un its ) LT O EI n c on v (a rb itr ar y un its ) LT O Δf _E In FF *t W t f ct Δf _E In w ei gh t Δf _E In av g, w ei gh t δ 85 67.8 100.0 -0.32 2.488 100 71.50738 100 -0.28 0.961 nvPM N T63 JP-8 HEFA TDyn 50 22 7 38.25739 100 -0.62 3.128 -0.48 30 47.10700 100 -0.53 1.48 85 65.3 100.0 -0.35 2.488 100 68.34655 100 -0.32 0.961 nvPM N T63 JP-8 FT GTLSyn 50 22 7 35.43809 100 -0.65 3.128 -0.51 30 42.68464 100 -0.57 1.48 85 62.7 100.0 -0.37 2.488 100 67.65956 100 -0.32 0.961 nvPM N T63 JP-8 HEFA C 50 22 7 35.78425 100 -0.64 3.128 -0.51 30 45.08090 100 -0.55 1.48 85 62.7 100.0 -0.37 2.488 100 65.18571 100 -0.35 0.961 nvPM N T63 JP-8 HEFA T 50 22 7 30.55271 100 -0.69 3.128 -0.53 30 41.73004 100 -0.58 1.48 85 62.7 100.0 -0.37 2.488 100 66.42264 100 -0.34 0.961 nvPM N CFM56 -2 JP-8 FT GTL 10 0 6 7 1.05E+14 1.17E+15 -0.91 3.128 -0.83 -0.66 0.33 30 1.30E+13 9.89E+14 -0.99 1.48 85 3.98E+14 1.45E+15 -0.72 2.488 100 5.39E+14 1.37E+15 -0.61 0.961 nvPM N GE CF700- 2-D-2 Jet A1 HEFA- SPK 10 0 15 7 5.70E+15 7.30E+15 -0.22 3.128 -0.17 30 6.08E+15 7.77E+15 -0.22 1.48 85 7.63E+15 9.07E+15 -0.16 2.488 100 9.83E+15 9.87E+15 0.00 0.961 nvPM N GE CF700- 2-D-2 Jet A1 FT- SPK 10 0 15 7 2.00E+14 7.30E+15 -0.97 3.128 -0.90 30 2.95E+14 7.77E+15 -0.96 1.48 85 1.27E+15 9.07E+15 -0.86 2.488 100 3.57E+15 9.87E+15 -0.64 0.961 nvPM N CFM56 -7B Jet A1 FT GTL 10 0 36 7 1.41E+00 1.00E+02 -0.99 3.128 -0.72 30 1.27E+01 1.00E+02 -0.87 1.48 85 5.77E+01 1.00E+02 -0.42 2.488 100 6.06E+01 1.00E+02 -0.39 0.961

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 18 An uncertainty weighted least squares quadratic fit (Δf_EIn vs. blend%) was performed using the data given in Table 5. The resulting values for the fit (Δf_EIn_fit) are given in Table 5. The fit function is: Δf_EIn_fit = 2.0𝐸 − 30 – 1.25𝐸 − 2 ∗ 𝑏𝑙𝑒𝑛𝑑% + 5.91𝐸 − 5 ∗ 𝑏𝑙𝑒𝑛𝑑%^2. (14) A quadratic function was found to work better than the linear expression given above in the introduction. Table 5: nvPM impact factor analysis blend% Δfavg, weight δ Δf_EIn 0 0 0 0.00 50 -0.48 0.10 -0.48 100 -0.66 0.33 -0.66 STEP 5 – The original Δf_EIn data (points) and the weighted quadratic function fit (dotted line) are shown in Figure 2. The blue data points represent the original impact factors, the orange data points represent the uncertainty weighted functional fit values. Figure 2: Δf_EIn data and uncertainty weighted quadratic function fit The original and fitted data along with an impact factor ER is given in -1.2 -1 -0.8 -0.6 -0.4 -0.2 0 0.2 -20 0 20 40 60 80 100 120 Δf _E In blend% Impact factor vs blend%

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 19 Table 6. The number-based emission index impact factor ER is defined as ERΔf_EIn = | Δf_EIn_fit - Δf_EIn |( ∗ ) (15)

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 20 Table 6: Δf_EInavg, weight, Δf_EInfit, and associated ER blend% Δfavg, weight δ Δf_EIn ER 0 0 0 0 0 50 -0.48 0.10 -0.48 0.01 100 -0.66 0.33 -0.66 0.01 The fact that the curve in Figure 2 passes through the data and their associated ERs are always less than unity, reveals that the Δf_EInfit function is a good representation for the EIn impact factor data and is thereby validated. The fit coefficients (constant, linear term, quadratic term) in the uncertainty weighted EIn impact factor equation contain uncertainty. Taking the uncertainty terms to add in quadrature, the uncertainty in the EIn impact factor (given above Table 4) function becomes: δΔf_EIn_fit = {(1.0𝐸 − 8)2 + (5.23𝐸 − 3 ∗ 𝑏𝑙𝑒𝑛𝑑%)2 + (7.73𝐸 − 5 ∗ 𝑏𝑙𝑒𝑛𝑑%2)2}1/2 (16) Illustrative example: To illustrate the use of the impact factor for nvPM N, assume an airport has normal particle number emissions of 1016/year. Assuming 12% of the jet fuel used at the airport is blended conventional/SAJF, at a 50% blend (50% blend  Δf_EIn = -0.48) the nvPM number emissions savings would be 5.73 x 1014/year (1016 * 0.12 * 0.48) with an uncertainty of 3.9x1014/year (1016*0.12*δΔf=0.325). 4.3 nvPM Mass For nvPM mass we anticipate a similar relationship as that for nvPM N. At currently approved blend percentages of up to 50%, the reduction in nvPM mass emissions is directly related to the amount of SAJF burned, thus the emission index of the blend (EImPM, blend), for a given engine and operating condition, can be expressed as follows EImPM,blend = 𝛼 ∗ EImPM,conv ∗ 1 − 𝑏𝑙𝑒𝑛𝑑%100% + 𝛽 ∗ EImPM,SAJF ∗ 𝑏𝑙𝑒𝑛𝑑%100% , (17) where EImPM, conv and EImPM, SAJF denote the mass-based emission indices for the conventional and alternate fuels. nvPM N Findings The fact that the uncertainty in Δf_EIn_fit is smaller than the absolute value of Δf_EIn_fit implies that there is a statistically meaningful nvPM N impact associated with alternative fuel usage. 𝛥f_EIn_fit = − 1.25𝐸 − 2 ∗ 𝑏𝑙𝑒𝑛𝑑% + 5.91𝐸 − 5 ∗𝑏𝑙𝑒𝑛𝑑%^2. δΔf_EIn_fit = { (5.23𝐸 − 3 ∗ 𝑏𝑙𝑒𝑛𝑑%)2 + (7.73𝐸 − 5∗ 𝑏𝑙𝑒𝑛𝑑%2)2}1/2

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 21 The impact on nvPM mass due to the SAJF for specific engines and their operating conditions is Δf_EIm = 𝐸ImPM,blend - EImPM,convEImPM,conv = 𝛼 ∗ 1 − 𝑏𝑙𝑒𝑛𝑑%100% + 𝛽 ∗ EImPM,ASJFEImPM,conv ∗ 𝑏𝑙𝑒𝑛𝑑%100% – 1 = 𝜑0 + 𝜑1 𝑏𝑙𝑒𝑛𝑑%100% 𝑤𝑖𝑡ℎ 𝜑0 = 𝛼 − 1, 𝜑1 = 𝛽 ∗ EImPM,ASJFEImPM,conv − 𝛼 . (18) STEP 1 – Based on the limited available data from the State of the Industry Report, the observational data is found to be dependent on power and blend%. For use in the AEDT model, an average value for impact factor, weighted over LTO cycle fuel burns, is calculated, resulting in impact factors depending on blend% alone. STEPS 2 AND 3 – The nvPM Mass emissions impact spreadsheet is given in Table 7 below. In the spreadsheet Δf_EIm denotes the power dependent nvPM Mass impact factors; FF*t denotes the product of fuel flow rate for a given operational mode and the time in mode as defined by the ICAO LTO cycle. This represents the fuel burned in the mode and is used as a weighting function to get an LTO weighted impact factor, Δf_EImweight. Δf_EImavg, weight is the average of all Δf_EIm weight values recorded for a given blend%. δ denotes the standard deviation in Δf_EImweight values and is used as the uncertainty in Δf_EImavg, weight.

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 22 Table 7: nvPM mass emissions data STEP 4 – Table 8 gives Δf_EImavg, weight for three values of blend%, with associated uncertainty. Δf_EIm is zero at blend%=0 with zero uncertainty, since the impact of alternate fuel is zero when there is no alternate fuel. Sp ec ie s En gi ne Co nv fu el SA JF bl en d% Re f # LT O pw r LT O EI n b le nd (a rb itr ar y un its ) LT O EI n c on v (a rb itr ar y un its ) LT O Δf _E Im FF *t W t f ct Δf _E Im w ei gh t Δf _E Im av g, w ei gh t δ nvPM M CFM56- 2C JP- 8 FT GTL 50 6 7 1.91E+00 1.01E+01 -0.81 3.1278 -0.77 -0.65 0.12 30 1.26E+00 1.40E+01 -0.91 1.48 85 2.89E+01 9.14E+01 -0.68 2.4882 100 4.46E+01 1.15E+02 -0.61 0.9611 nvPM M GE CF700- 2-D-2 Jet A1 HEFA- SPK 50 15 7 8.52E+00 1.70E+01 -0.50 3.1278 -0.66 30 9.417808 28.68151 -0.67 1.48 85 23.67424 114.5833 -0.79 2.4882 100 60.60606 296.4015 -0.80 0.9611 nvPM M CFM56- 7B Jet A1 FT GTL 50 36 7 3.52E+01 1.00E+02 -0.65 3.1278 -0.53 30 3.66E+01 1.00E+02 -0.63 1.48 85 6.34E+01 1.00E+02 -0.37 2.4882 100 6.06E+01 1.00E+02 -0.39 0.9611 nvPM M CFM56- 2C JP- 8 FT GTL 100 6 7 5.66E-01 1.01E+01 -0.94 3.1278 -0.93 -0.70 0.24 30 1.17E-01 1.40E+01 -0.99 1.48 85 8.77E+00 9.14E+01 -0.90 2.4882 100 1.80E+01 1.15E+02 -0.84 0.9611 nvPM M GE CF700- 2-D-2 Jet A1 HEFA- SPK 100 15 7 1.14E+01 1.70E+01 -0.33 3.1278 -0.38 30 16.73412 28.68151 -0.42 1.48 85 67.23485 114.5833 -0.41 2.4882 100 183.7121 296.4015 -0.38 0.9611 nvPM M GE CF700- 2-D-2 Jet A1 FT- SPK 100 15 7 5.68E+00 1.70E+01 -0.67 3.1278 -0.81 30 5.681818 28.68151 -0.80 1.48 85 7.575758 114.5833 -0.93 2.4882 100 13.25758 296.4015 -0.96 0.9611 nvPM M CFM56- 7B Jet A1 FT GTL 100 36 7 3.24E+01 1.00E+02 -0.68 3.1278 -0.69 30 9.86E+00 1.00E+02 -0.90 1.48 85 3.94E+01 1.00E+02 -0.61 2.4882 100 3.66E+01 1.00E+02 -0.63 0.9611

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 23 An uncertainty weighted least squares quadratic fit is performed using the uncertainties given in Table 8. The resulting values for the fit (Δf_EIm_fit) are given in Table 8. The fit function is: Δf_EIm_fit = −7.98𝐸 − 31 – 1.90𝐸 − 2 ∗ 𝑏𝑙𝑒𝑛𝑑% + 1.20𝐸 − 4 ∗ 𝑏𝑙𝑒𝑛𝑑%^2. (19) A quadratic function was found to work better than the linear expression given above in the introduction. Table 8: nvPM mass impact factor analysis blend% Δfavg, weight δ Δf_EImfit 0 0 0 0.00 50 -0.65 0.12 -0.65 100 -0.70 0.24 -0.70 STEP 5 – The original Δf_EIm data (orange points) and the weighted quadratic function fit (dotted line) are shown in Figure 3. Figure 3: Δf_EIm data and uncertainty weighted quadratic function fit The original and fitted data along with an impact factor ER is given in Table 9. The ER is defined as ERΔf_EIn = |Δf_EIn_fit - Δf_EIn| ∗ (20) -1 -0.8 -0.6 -0.4 -0.2 0 0.2 -20 0 20 40 60 80 100 120 Δf _E Im blend% Impact factor vs blend%

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 24 Table 9: Δfavg, weight, Δf_EImfit, and associated ER blend% Δfavg, weight δ Δf_EImfit ER 50 -0.65 0.12 -0.65 0.00 100 -0.70 0.24 -0.70 0.00 The fact that the fitted curve passes through the data error bars in Figure 3 and their associated ERs are always less than unity, reveals that the Δf_EImfit function is a good representation for the EIm impact factor data and is thereby validated. The fit coefficients (constant, linear term, quadratic term) in the uncertainty weighted EIm impact factor equation contain uncertainty. Taking the uncertainty terms to add in quadrature, the uncertainty in the EIm impact factor (given above Table 8) function becomes: 𝛿𝛥𝑓_𝐸𝐼𝑚_𝑓𝑖𝑡 = {(1.0𝐸 − 8)2 + (5.31𝐸 − 3 ∗ 𝑏𝑙𝑒𝑛𝑑%)2 + (6.70𝐸 − 5 ∗ 𝑏𝑙𝑒𝑛𝑑%2) 2 }1/2 (21) Illustrative example: To illustrate the use of the impact factor for nvPM Mass, assume an airport has normal particle mass emissions of 1000 kg/year. Assuming 12% of the jet fuel used at the airport is blended conventional/SAJF, at a 50% blend (50% blend  Δf_EIm = -0.65) the nvPM mass emissions savings would be 78 kg/year (1000 * 0.12 * 0.65) with an uncertainty of 38 kg/year (1000*0.12*δΔf=0.314). 4.4 NOx For NOx we anticipated a small if not negligible impact. At currently approved blend percentages of up to 50%, the reduction in NOx emissions is directly related to the amount of SAJF burned, thus the emission index of the blend (EINOx, blend), for a given engine and operating condition, can be expressed in a similar manner as used for SOx, nvPM N, and nvPM Mass: EINOx,blend = 𝛼 ∗ EINOx,conv ∗ 1 − 𝑏𝑙𝑒𝑛𝑑%100% + 𝛽 ∗ EINOx,SAJF ∗ 𝑏𝑙𝑒𝑛𝑑%100% , (22) nvPM Mass Findings The fact that the uncertainty in Δf_EIm_fit is smaller than the absolute value of Δf_EIm_fit implies that there is a statistically meaningful nvPM M impact associated with alternative fuel usage. Δf_EIm_fit = − 1.90𝐸 − 2 ∗ 𝑏𝑙𝑒𝑛𝑑% + 1.20𝐸 − 4 ∗ 𝑏𝑙𝑒𝑛𝑑%^2 𝛿Δf_EIm_fit = { (5.31𝐸 − 3 ∗ 𝑏𝑙𝑒𝑛𝑑%)2 + (6.70𝐸 − 5∗ 𝑏𝑙𝑒𝑛𝑑%^2) 2}1/2

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 25 The impact on NOx emissions due to the SAJF for specific engines and their operating conditions is 𝛥f_EINOx = EINOx,blend EINOx,convEINOx,conv = 𝛼 ∗ 1 − 𝑏𝑙𝑒𝑛𝑑%100% + 𝛽 ∗ EINOx,ASJFEINOx,conv ∗ 𝑏𝑙𝑒𝑛𝑑%100% – 1 = 𝜑0 + 𝜑1 ∗𝑏𝑙𝑒𝑛𝑑%100% (𝑤𝑖𝑡ℎ 𝜑0 = 𝛼 − 1, 𝜑1 = 𝛽 ∗ EINOx,ASJFINOx,con − 𝛼). (23) STEP 1 – Based on the limited available data from the State of the Industry Report, the observational data is found to be dependent on power and blend%. For use in the AEDT model, an average value for impact factor, weighted over LTO cycle fuel burns, is calculated, resulting in impact factors depending on blend% alone. STEP 2 AND 3 – The NOx impact spreadsheet is given in Table 10 below. In the spreadsheet Δf_EINOx denotes the NOx impact factors after averaging over the ICAO LTO cycle. Δf_EINOx, avg is the average of all Δf_EINOx values recorded for a given blend%. δ denotes the standard deviation in Δf_EINOx values and is used as the uncertainty in Δf_EINOx, avg. Table 10: NOx emissions impact spreadsheet Sp ec ie s En gi ne ty pe Co nv fu el SA JF bl en d% Re f # Δf _E IN Ox Δf _E IN Ox _A v g δ NOx SaM146 Jet A1 DSHC 10 42 0 -0.0253 0.04376 NOx GTCP85 Garret Honeywell APU Jet A1 UCO SPK 10 17 0 NOx CFM56-5C4 Jet A1 DSHC 10 42 -0.0758 NOx SaM146 Jet A1 DSHC 20 42 0 -0.0783 0.12378 NOx GTCP85 Garret Honeywell APU Jet A1 UCO SPK 20 17 0 NOx CFM56-5C4 Jet A1 DSHC 20 42 -0.0532 NOx GTCP85 Garret Honeywell APU Jet A1 UCO SPK 25 17 0 -0.0033 0.00577 NOx CFM56-7B Jet A Bio-SPK 25 51 -0.01 NOx GTCP85 Garret Honeywell APU Jet A1 UCO SPK 40 17 0 -0.0899 0.1557 NOx 9 pt. lean direct low emissions combustor JP-8 AMJ 50 49 0 -0.0047 0.01643 NOx AE 3007 combustor JP-8 ATJH SPK 50 26 0 NOx TFE34 JP-8 ATJH SPK 50 26 0 NOx PW615F JP-8 ATJH SPK 50 26 0 NOx TPE331-10YGD JP-8 ATJH SPK 50 26 0 NOx F117-PW-100 = PW2000 JP-8 Beef Tallow 50 20 0 NOx T63-A-701 JP-8 Beef Tallow 50 21 0 NOx TPE331-10 JP-8 Bio-SPK 50 10 0

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 26 Sp ec ie s En gi ne ty pe Co nv fu el SA JF bl en d% Re f # Δf _E IN Ox Δf _E IN Ox _A v g δ NOx T63-A-703 JP-8 Fats & Grease 50 21 0 NOx T63-A-700 JP-8 FT GTL 50 21 0 NOx CFM56-7 JP-8 FT GTL 50 22 0 NOx CFM56-2 JP-8 FT GTL 50 22 0 NOx F117 JP-8 FT GTL 50 22 0 NOx TF33 JP-8 FT GTL 50 22 0 NOx PW308 JP-8 FT GTL 50 22 0 NOx T63-A-700 JP-8 HEFA 50 21 0 NOx CFM56-7 JP-8 HEFA 50 22 0 NOx CFM56-2 JP-8 HEFA 50 22 0 NOx F117 JP-8 HEFA 50 22 0 NOx TF33 JP-8 HEFA 50 22 0 NOx PW308 JP-8 HEFA 50 22 0 NOx JT9D-7R4G2 Jet A HVO 50 51 0 NOx GTCP85 Garret Honeywell APU Jet A1 UCO SPK 50 17 0 NOx CFM56-7B Jet A Bio-SPK 50 51 -0.05 NOx CFM56-7 Jet A1 FT-GTL 50 47 -0.0673 NOx T63-A-701 JP-8 Beef Tallow 100 21 0 -0.032 0.07233 NOx T63-A-703 JP-8 Fats & Grease 100 21 0 NOx T63-A-700 JP-8 FT GTL 100 21 0 NOx T63-A-700 JP-8 HEFA 100 21 0 NOx GTCP85 Garret Honeywell APU Jet A1 UCO SPK 100 17 0 NOx CFM56-7 Jet A1 FT-GTL 100 47 -0.1941 NOx MK113 APU Artouste Jet A1 FT-GTL 100 34 -0.030 STEP 4 –Table 11 takes selected parameters from Table 10 for further analysis. Table 11: NOx impact factors vs. blend% blend% Δf_EINOx_Avg δ 10 -0.0253 0.04376 20 -0.018 0.031 25 -0.0033 0.00577 40 0 0.006

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 27 blend% Δf_EINOx_Avg δ 50 -0.0047 0.01643 100 -0.032 0.07233 Figure 4 shows a plot of Δf_EINOx, Avg vs blend% with associated uncertainties. This plot suggests that a constant function is the best fit to the data. An uncertainty weighted least squares fit of a constant to the data was performed yielding a result of Δf_EINOx = −0.0024 ± 0.0039 (24) The fact that the uncertainty in Δf_EINOx (0.0039) is greater than the absolute value of Δf_EINOx (0.0024) implies that there is no statistically meaningful NOx impact associated with alternative fuel usage. Figure 4: Δf_EINOx, Avg vs blend% and associated uncertainties STEP 5 – The original Δf_EINOx data and the weighted constant function fit values are shown in Figure 5. The blue data points represent the original impact factors, the orange data points represent the uncertainty weighted functional fit values. -0.12 -0.1 -0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0 20 40 60 80 100 Δf _E IN O Blend% NOx Impact

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 28 Figure 5: NOx impacts vs. blend%. Blue circles represent original data; orange line gives functional fit. The original and fitted data along with an impact factor ER is given in Table 12. The ER is defined as ERΔf_EINOx = Δf_EINOx_fit - Δf_EINOx(δ + δfit) (25) with δfit = 0.0039 denotes the standard deviation in the constant fit. Table 12: Δf_EINOx-Avg, Δf_EINOx, fit and associated ER blend% Δf_EINOx_Avg δ Δf_EINOx_fit δ_fit ER 10 -0.0253 0.04376 -0.0024 0.0039 0.48 20 -0.018 0.031 -0.0024 0.0039 0.44 25 -0.0033 0.00577 -0.0024 0.0039 0.10 40 0 0.006 -0.0024 0.0039 -0.25 50 -0.0047 0.01643 -0.0024 0.0039 0.11 100 -0.032 0.07233 -0.0024 0.0039 0.39 -0.12 -0.1 -0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0 20 40 60 80 100 Δf _E IN O Blend% NOx Impact

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 29 The fact that the error bars in Figure 5 overlap and their associated ERs are always less than unity, reveals that the Δf_EINOx function is a good representation for the EINOx impact factor data and is thereby validated. Illustrative example: To illustrate the use of the impact factor for NOx, assume an airport has normal NOx emissions of 1000 kg/year. Assuming 12% of the jet fuel used at the airport is blended conventional/SAJF, at a 50% blend the NOx emissions savings would be 0.29 kg/year (with an uncertainty of 0.47 kg/year). The fact that the uncertainty in the NOx emissions savings (0.47 kg/year) is greater than the absolute value of savings (0.29 kg/year) implies that there is no statistically meaningful NOx impact associated with alternative fuel usage. 4.5 CO For CO we anticipated a similar relationship as that for nvPM. At currently approved blend percentages of up to 50%, the reduction in CO emissions is directly related to the amount of SAJF burned, thus the emission index of the blend (EICO, blend), for a given engine and operating condition, can be expressed in a similar manner as used for SOx, nvPM N, and nvPM Mass: EICO,blend = 𝛼 ∗ EinCO,conv ∗ 1 − 𝑏𝑙𝑒𝑛𝑑%100% + 𝛽 ∗ EinCO,SAJF ∗ 𝑏𝑙𝑒𝑛𝑑%100% , (26) The impact on CO emissions due to the SAJF for specific engines and their operating conditions is Δf_EICO = (EICO,blend – EICO,conv) EICO,conv = 𝛼 ∗ 1 − %% + 𝛽 ∗ EICO,SAJFEICO,conv ∗ %% – 1 = 𝜑0 + 𝜑1 ∗ %% 𝑤𝑖𝑡ℎ 𝜑0 = 𝛼 − 1, 𝜑1 = 𝛽 ∗ EICO,SAJFEICO,conv − 𝛼 . (27) STEP 1 – Based on the limited available data from the State of the Industry Report, all the observational data is found to be dependent on blend%. For those cases where an engine power dependency was reported, a weighted average was used, where the weight function was the fuel burned for the ICAO LTO cycle, similar to the NOx analysis. STEP 2 AND 3 – The CO impact spreadsheet is given in Table 13 below. In the spreadsheet Δf_EICO denotes the CO impact factors, and Δf_EICO, avg is the average of all Δf_EICO values recorded for a given blend%. δ denotes the standard deviation in Δf_EICO values and is used as the uncertainty in Δf_EICO, avg. NOx Findings The fact that the uncertainty in Δf_EINOx (0.0039) is greater than the absolute value of Δf_EINOx (0.0024) implies that there is no statistically meaningful NOx impact associated with alternative fuel usage.

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 30 Table 13: CO emissions impact spreadsheet Species Engine Conv fuel SAJF blend% Ref # Δf_EI CO Δf_EI CO_Avg δ CO Combustor JP8 FT 50 44 -0.242 -0.129 0.119 CO PW2000 JP8 HRJ 50 20 -0.265 CO T63 JP8 FT shell 50 21 -0.087 CO CFM56-7 Jet A1 FT 50 47 0.007 CO T701-C JP8/JetA1 FT & HEFA 50 22 -0.059 CO Combustor JP8 FT 100 44 -0.402 -0.174 0.106 CO T63 JP8 FT shell 100 21 -0.258 CO T63 JP8 FT Rentech 100 21 -0.111 CO T63 JP8 HRJ R8 100 21 -0.129 CO T63 JP8 HRJ tallow 100 21 -0.114 CO T63 JP8 HRJ Came 100 21 -0.154 CO CFM56-7 Jet A1 FT 100 47 -0.085 CO T701-C JP8/JetA1 FT shell 100 22 -0.140 STEP 4 – Table 14 takes selected parameters from Table 13 for further analysis. Table 14: CO impact factors for selected blend%, with uncertainty blend% Δf EI CO δ 0 0 0 50 -0.129 0.0660 100 -0.174 0.1319

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 31 Figure 6: CO impact vs. blend% Figure 6 shows a plot of Δf_EICO, Avg vs blend% with associated uncertainties. This plot suggests that a linear function is the best fit to the data. An uncertainty weighted linear least squares fit to the data was performed yielding a result of: Δf_EICO = −2.41𝐸 − 16 – 2.16𝐸 − 3 ∗ 𝐵𝑙𝑒𝑛𝑑% (28) STEP 5 – The original Δf_EICO data and the weighted linear function fit values are shown in Figure 7. The blue data points represent the original impact factors, the orange data points represent the uncertainty weighted functional fit values. -0.35 -0.3 -0.25 -0.2 -0.15 -0.1 -0.05 0 -20 0 20 40 60 80 100 120 Δf _C O Blend% CO Impact

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 32 Figure 7: CO impact vs. blend% with linear fit The original and fitted data along with an impact factor ER is given in Table 15. The ER is defined as ERΔf_EICO = | f_EICO_fit f_EICO |( ) . (29) Table 15: Δf_EICO_Avg, Δf_EICO_fit and associated ER blend% Δf EI CO δ Δf EI COFit ER 0 0 0 2.41E-16 0 50 -0.129 0.0660 -0.108 0.16 100 -0.174 0.1319 -0.216 0.16 The fact that the error bars in Figure 7 overlap and their associated ERs are always less than unity, reveals that the Δf_EICO function is a good representation for the EICO impact factor data and is thereby validated. -0.4 -0.35 -0.3 -0.25 -0.2 -0.15 -0.1 -0.05 0 0.05 -20 0 20 40 60 80 100 120 Δf _C O Blend% CO Impact

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 33 The fit coefficients (constant, linear term) in the uncertainty weighted CO impact factor equation contain uncertainty. Taking the uncertainty terms to add in quadrature, the uncertainty in the CO impact factor function (given above step 5) becomes: δΔfCO, fit = {(1.0E-8)2 + (9.32E-4*blend%)2 }1/2 (30) Illustrative example: To illustrate the use of the impact factor for CO, assume an airport has normal CO emissions of 1000 kg/year. Assuming 12% of the jet fuel used at the airport is blended conventional/SAJF, at a 50% blend (50% blend  Δf = -0.108) the CO emissions savings would be 13 kg/year (1000*0.12*0.108) with an uncertainty of 5.6 kg/year (1000*0.12*δΔf=0.047). 4.6 UHC For UHC we anticipated a similar relationship as that for CO. At currently approved blend percentages of up to 50%, the reduction in UHC emissions is directly related to the amount of SAJF burned, thus the emission index of the blend (EIUHC, blend), for a given engine and operating condition, can be expressed in a similar manner as used for SOx, nvPM N, and nvPM Mass: EIUHC,blend = 𝛼 ∗ 𝐸𝐼UHC,conv ∗ 1 − 𝑏𝑙𝑒𝑛𝑑%100% + 𝛽 ∗ 𝐸IUHC,SAJF ∗ 𝑏𝑙𝑒𝑛𝑑%100% , (31) The impact on UHC emissions due to the SAJF for specific engines and their operating conditions is Δf_EIUHC = UHC,blend – UHC,convUHC,conv = 𝛼 ∗ 1 − %% + 𝛽 ∗ UHC,SAJFUHC,conv ∗ %% – 1 = 𝜑0 + 𝜑1 ∗ %% 𝑤𝑖𝑡ℎ 𝜑0 = 𝛼 − 1, 𝜑1 = 𝛽 ∗ EIUHC,SAJFEIUHC,conv − 𝛼 . (32) STEP 1 – Based on the limited available data from the State of the Industry Review, all the observational data is found to be dependent on blend%. For those cases where an engine power dependency was reported, a weighted average was used, where the weight function was the fuel burned for the ICAO LTO cycle, similar to the NOx and CO analyses. CO Findings The fact that the uncertainty in Δf_EICO is smaller than the absolute value of Δf_EICO implies that there is a statistically meaningful CO impact associated with alternative fuel usage. Δf_EICO = – 2.16E-3*Blend% δΔfCO, fit = 9.32E-4*blend%

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 34 STEP 2 AND STEP 3 – The UHC impact spreadsheet is given in Table 16 below. In the spreadsheet Δf_EIUHC denotes the UHC impact factors, and Δf_EIUHC, avg is the average of all Δf_EIUHC values recorded for blend%s of 50% and 100%. For 25% blend%, the average is taken over blend%s from 5% to 40%; for 75% blend%, the average is taken over blend%s from 60% to 95%. δ denotes the standard deviation in Δf_EIUHC values over the range of blend% used for the average, and is used as the uncertainty in Δf_EIUHC, avg. Table 16: UHC spreadsheet Species Conv fuel SAJF blend% Ref # Δf_EI UHC Δf_EI UHC_Avg δ UHC Jet A1 HEFA 5 17 -0.527 UHC Jet A1 HEFA 10 17 -0.175 UHC Jet A1 HEFA 15 17 -0.438 UHC Jet A1 HEFA 20 17 -0.556 UHC Jet A1 HEFA 25 17 -0.525 -0.399 0.168 UHC Jet A1 HEFA 30 17 -0.155 UHC Jet A1 HEFA 40 17 -0.417 UHC JP-8 FT GTL 50 21 -0.027 -0.140 0.219 UHC JP-8 FT GTL 50 6 0.000 UHC Jet A1 HEFA 50 17 -0.393 UHC Jet A1 HEFA 60 17 -0.417 UHC Jet A1 HEFA 70 17 -0.382 UHC Jet A1 HEFA 75 17 -0.510 -0.447 0.053 UHC Jet A1 HEFA 80 17 -0.418 UHC Jet A1 HEFA 85 17 -0.467 UHC Jet A1 HEFA 90 17 -0.413 UHC Jet A1 HEFA 95 17 -0.523 UHC JP-8 FT GTL 100 21 -0.264 -0.257 0.053 UHC JP-8 FT GTL 100 21 -0.253 UHC JP-8 HRJ R8 100 21 -0.246 UHC JP-8 HRJ tallow 100 21 -0.235 UHC JP-8 HRJ Came 100 21 -0.348 UHC JP-8 FT GTL 100 6 -0.171 UHC Jet A1 HEFA 100 17 -0.282 STEP 4 –

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 35 Table 17 takes selected parameters from Table 16 for further analysis.

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 36 Table 17: UHC impact factors for selected blend%s, with uncertainty blend% Δf_EI UHC_Avg δ 0 0 0 25 -0.399 0.168 50 -0.140 0.219 75 -0.447 0.053 100 -0.257 0.053 Figure 8: UHC impact vs. blend% Figure 8 shows a plot of Δf_EIUHC, Avg vs blend% with associated uncertainties. This plot suggests a function that is 0 at 0 blend percent, decreases as blend% increases initially, and asymptotically reaches a constant value at larger blend%s. A best least square fit to this data was done with a hyperbolic tangent function, which has an appropriate shape. The result is given by Δf_EIUHC = −0.3482 ∗ 𝑡𝑎𝑛ℎ (0.322 ∗ 𝑏𝑙𝑒𝑛𝑑%). (34) A hyperbolic tangent function was also used to fit the uncertainty (δ) in Δf_EIUHC as shown in Figure 9, with the result δΔf_EIUHC = 0.1234 ∗ 𝑡𝑎𝑛ℎ (0.2867 ∗ 𝑏𝑙𝑒𝑛𝑑%). (35) STEP 5 – The original Δf_EIUHC data and the uncertainty weighted function fit values are shown in Figure 9. The blue data points represent the original impact factors, the orange data points represent the uncertainty weighted functional fit values. -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 -20 0 20 40 60 80 100 120 Δf _E I U HC Blend% UHC Impact factor vs blend%

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 37 Figure 9: UHC impact vs. blend% with its functional fit The original and fitted data along with an impact factor ER is given in Table 18. The ER is defined as ERΔf_EIUHC = | Δf_EIUHC_fit - Δf_EIUHC |( _ ) (36) Table 18: Δf_EIUHC_Avg, Δf_EIUHC_fit and associated ER blend% Δf EI UHC δ Δf EI UHCfit δΔf_EIUHC ERΔf_EIUHC 0 0 0.0053 0.0000 0.0053 0.00 25 -0.3990 0.1677 -0.3482 0.1234 0.17 50 -0.1400 0.2195 -0.3482 0.1234 -0.61 75 -0.4471 0.0534 -0.3482 0.1234 0.56 100 -0.2570 0.0533 -0.3482 0.1234 -0.52 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 -20 0 20 40 60 80 100 120 Δf _E I U HC Blend% UHC Impact factor vs blend%

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 38 The fact that the error bars in Figure 9 overlap and their associated ERs are always less than unity, reveals that the Δf_EIUHC function is a good representation for the EIUHC impact factor data and is thereby validated. Illustrative example: To illustrate the use of the impact factor for UHC, assume an airport has normal UHC emissions of 1000 kg/year. Assuming 12% of the jet fuel used at the airport is blended conventional/SAJF, at a 50% blend (50% blend  Δf = -0.348) the UHC emissions savings would be 41.8 kg/year (1000 * 0.12 * 0.348) with an uncertainty of 14.8 kg/year (1000*0.12*δΔf=0.123). Note: The functional fit analysis for UHC impact is confounded by the extensive scatter in the small amount of data available in the literature on UHC emissions (three papers (Refs 6, 17, 21) with 80% of the data coming from Ref 17). The observed scatter appears to be driven by not only blend ratio but also engine operating condition (Ref 17). As a result, the authors caution applying the impact factors for UHC resulting from the above functional analysis. The authors further recommend the pursuit of additional experimental studies on UHC emissions associated with blended SAJFs as a function of engine operating condition in order to strengthen confidence in the resulting impact factor analysis. 4.7 HAPs For HAPs we anticipated a small if not negligible impact. At currently approved blend percentages of up to 50%, the reduction in HAPs emissions is directly related to the amount of SAJF burned, thus the emission index of the blend (EInHAPs), for a given engine and operating condition, can be expressed in a similar manner as used for SOx, nvPM N, and nvPM Mass: 𝐸𝐼HAPs,blend = 𝛼 ∗ 𝐸𝐼HAPs,conv ∗ 1 − %% + 𝛽 ∗ 𝐸𝐼HAPs,SAJF ∗ %% , (37) The impact on HAPs emissions due to the SAJF for specific engines and their operating conditions is Δf_EIHAPs = EIHAPs,blend – EIHAPs,conv EIHAPs,conv = 𝛼 ∗ 1 − %% + 𝛽 ∗ EIHAPs,ASJFEIHAPs,conv ∗ %% – 1 UHC Findings The fact that the uncertainty in Δf_EIUHC is smaller than the absolute value of Δf_EIUHC implies that there may be a statistically meaningful UHC impact associated with alternative fuel usage. Due to the variability of available data, the team was not able to definitively derive a relationship between blend percentage and UHC emissions reduction that can be confidently used to describe UHC emissions. Δf_EIUHC = −0.3482 ∗ 𝑡𝑎𝑛ℎ (0.322 ∗ 𝑏𝑙𝑒𝑛𝑑%). δΔf EIUHC = 0.1234 ∗ 𝑡𝑎𝑛ℎ (0.2867 ∗ 𝑏𝑙𝑒𝑛𝑑%).

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 39 = 𝜑0 + 𝜑1 ∗ %% 𝑤𝑖𝑡ℎ 𝜑0 = 𝛼 − 1, 𝜑1 = 𝛽 ∗ EIHAPs,ASJFEIHAPs,conv − 𝛼 . (38) STEP 1 – Based on the limited available data from the State of the Industry Report, the observational data is found to be dependent on power and blend%. For use in the AEDT model, an average value for impact factor, weighted over LTO cycle fuel burns, is calculated, resulting in impact factors depending on blend% alone. STEP 2 AND STEP 3 – The HAPs impact spreadsheet is given in Table 19 below. In the spreadsheet Δf_EIHAPs denotes the HAPs impact factors after averaging over the ICAO LTO cycle. Δf_EIHAPs, avg is the average of all Δf_EIHAPs values recorded for a given blend%. δ denotes the standard deviation in Δf_EIHAPs values and is used as the uncertainty in Δf_EIHAPs, avg. Table 19: HAPs spreadsheet Species Conv fuel SAJF blend% Ref # Δf_EI HAP Δf_EI HAP_Avg δ HAP HCHO JP-8 HRJ 50 20 -0.2789 -0.1590 0.2962 HAP CH3CHO JP-8 HRJ 50 20 -0.4278 HAP HCHO JP-8 HRJ+FT 50 20 -0.3823 HAP CH3CHO JP-8 HRJ+FT 50 20 -0.4049 HAP HCHO Jet A1 HEFA 50 35 0.0039 HAP CH3CHO Jet A1 HEFA 50 35 0.4599 HAP Acrolein Jet A1 HEFA 50 35 0.1142 HAP C6H6 JP-8 HRJ.C 50 22 -0.2316 HAP C7H8 JP-8 HRJ.C 50 22 -0.2831 HAP HCHO Jet A1 HEFA 75 35 -0.0115 0.1611 0.1731 HAP CH3CHO Jet A1 HEFA 75 35 0.3347 HAP Acrolein Jet A1 HEFA 75 35 0.1602 HAP HCHO Jet A1 FT GTL 100 47 -0.0378 -0.3091 0.2203 HAP HCHO Jet A1 FT GTL 100 35 -0.1400 HAP CH3CHO Jet A1 FT GTL 100 35 0.0254 HAP Acrolein Jet A1 FT GTL 100 35 -0.5069 HAP C6H6 JP-8 HRJ.C 100 22 -0.4124 HAP C6H6 JP-8 HRJ.T 100 22 -0.4407 HAP C7H8 JP-8 HRJ.C 100 22 -0.4752 HAP C7H8 JP-8 HRJ.T 100 22 -0.4853 STEP 4 – Table 20 takes selected parameters from Table 19 for further analysis.

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 40 Table 20: HAPs impact factors vs. blend% blend% Δf_EI HAP_Avg δ 0 0 0 50 -0.159 0.296 75 0.161 0.173 100 -0.309 0.220 Figure 10 shows a plot of Δf_EIHAPs, Avg vs blend% with associated uncertainties. This plot suggests that a constant function is the best fit to the data. An uncertainty weighted least squares fit of a constant to the data was performed yielding a result of Δf_EIHAP = -0.006 ± 0.046 (39) The fact that the uncertainty in Δf_EIHAP (0.046) is greater than the absolute value of Δf_EIHAP (0.006) implies that there is no statistically meaningful HAP impact associated with alternative fuel usage. Figure 10: HAPs impact factors vs. blend% STEP 5 – The original Δf_EIHAPs data and the weighted constant function fit values are shown in Figure 11. The blue data points represent the original impact factors, the orange data points represent the uncertainty weighted functional fit values. -1.5 -1 -0.5 0 0.5 1 1.5 -20 0 20 40 60 80 100 120Δf _E I H AP Blend% HAP Impact factor vs blend%

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 41 Figure 11: HAPs impact vs. blend% The original and fitted data along with an impact factor ER is given in Table 21. The ER is defined as ERΔf_EIHAPs = | Δf_EIHAPs_fit - Δf_EIHAPs |δ+ δfit,HAPs (40) with δfit, HAPs = 0.046 denotes the standard deviation in the constant fit. Table 21: Δf_EIHAPs_Avg, Δf_EIHAPs_fit and associated ER blend% Δf EI HAP δ Δf EI HAPfit ER 0 0 0 -0.0181 0 50 -0.1590 0.2962 -0.006 0.26 75 0.1611 0.1731 -0.006 -0.48 100 -0.3091 0.2203 -0.006 0.69 -1.5 -1 -0.5 0 0.5 1 1.5 -20 0 20 40 60 80 100 120Δf _E I H AP Blend% HAP Impact

Emissions Quantification Methodology Report: ACRP 02-80 Quantifying Emissions Reductions at Airports from the Use of Alternative Jet Fuel Emissions Quantification Methodology Report Page 42 The fact that the error bars in Figure 11 overlap and their associated ERs are always less than unity, reveals that the Δf_EIHAPs function is a good representation for the EIHAPs impact factor data and is thereby validated. Illustrative example: To illustrate the use of the impact factor for HAPs, assume an airport has normal HAPs emissions of 1000 kg/year. Assuming 12% of the jet fuel used at the airport is blended conventional/SAJF, at a 50% blend the HAPs emissions savings would be 2.2 kg/year (with an uncertainty of 5.3 kg/year). The fact that the uncertainty in the HAPs emissions savings (5.3 kg/year) is greater than the absolute value of savings (2.2 kg/year) implies that there is no statistically meaningful HAP impact associated with alternative fuel usage. HAPs Findings The fact that the uncertainty in Δf_EIHAP (0.046) is greater than the absolute value of Δf_EIHAP (0.006) implies that there is no statistically meaningful HAP impact associated with alternative fuel usage.