Targeted Use of Sustainable Aviation Fuel to Maximize Climate Benefits

Sustainable aviation fuel (SAF) can reduce aviation’s CO2 and non-CO2 impacts. We quantify the change in contrail properties and climate forcing in the North Atlantic resulting from different blending ratios of SAF and demonstrate that intelligently allocating the limited SAF supply could multiply its overall climate benefit by factors of 9–15. A fleetwide adoption of 100% SAF increases contrail occurrence (+5%), but lower nonvolatile particle emissions (−52%) reduce the annual mean contrail net radiative forcing (−44%), adding to climate gains from reduced life cycle CO2 emissions. However, in the short term, SAF supply will be constrained. SAF blended at a 1% ratio and uniformly distributed to all transatlantic flights would reduce both the annual contrail energy forcing (EFcontrail) and the total energy forcing (EFtotal, contrails + change in CO2 life cycle emissions) by ∼0.6%. Instead, targeting the same quantity of SAF at a 50% blend ratio to ∼2% of flights responsible for the most highly warming contrails reduces EFcontrail and EFtotal by ∼10 and ∼6%, respectively. Acknowledging forecasting uncertainties, SAF blended at lower ratios (10%) and distributed to more flights (∼9%) still reduces EFcontrail (∼5%) and EFtotal (∼3%). Both strategies deploy SAF on flights with engine particle emissions exceeding 1012 m–1, at night-time, and in winter.


S1
nvPM EIn reductions due to sustainable aviation fuels 30 Two different methodologies are available to estimate the change in nvPM EIn from fuels with 31 different hydrogen mass content (Hfuel). Brem  the SAF (HSAF) and "reference" Jet A-1 fuel (Href). However, Brem et al. 1 highlighted that Eq. 36 (S1) is only valid for ̂> 30% and Δ < 0.6, and Figure S1 confirms that an extrapolation 37 beyond these limits can lead to unrealistic values where ΔnvPM EI n < −100%. 38 Alternatively, the standardised fuel composition correction model was also developed for the 39 ICAO Committee on Aviation Environmental Protection (CAEP/11) to account for the 40 variability/differences in the properties of conventional kerosene fuel, so that the measured 41 nvPM EIn corresponds to a Hfuel of 13.8% (Appendix 6.2.2 of the ICAO Annex 16 Vol. II) 2 , 42 Corrected nvPM EI n = nvPM EI n × fuel_N , where fuel,N = exp {(0.99 00 − 1.05)(13.8 − fuel )}.
fuel,N is the fuel composition correction factor for the nvPM EIn and 00 is the engine thrust 43 settings (in decimals). Figure S2 shows the estimated ΔnvPM EI n from the ICAO CAEP/11 44 model across a range of ̂ and Δ relative to a Hfuel of 13.8%. We note that Eq.

56
In this study, we extend the methodology of Brem et al. 1 to account for its known limitations. 57 The extended methodology, outlined in Eq. (1) in the main text and visualised in Figure S3, 58 utilises latest measurements from the NASA ACCESS 3 and ECLIF II/ND-MAX 4-6 59 experimental campaigns, which measured the nvPM EIn emitted by SAF with higher Hfuel of 60

S6
We compare the differences in the estimated ΔnvPM EI n between the extended fuel 77 composition correction model (Eq. 1 in the main text) and the ICAO CAEP/11 (Eq. S2). Figure  78 S4a shows that arithmetic differences between the two models range between -5% and +20% 79 across the full range of ̂ and Δ . However, the arithmetic difference between the two models 80 reduces to between 0% and +8% when we constrain Δ for which the ICAO CAEP/11 method 81 is valid, i.e., Δ of between -0.4% and +0.5%, and to the range of ̂ that is typically used in 82 cruise conditions, i.e., between 39% (5 th percentile) and 78% (95 th percentile) ( Figure S4b). 83 The estimated ΔnvPM EI n from both models are also compared against: (i) ground 84 measurements from the A-PRIDE 1 , EMPAIREX 7 , and ECLIF2/ND-MAX 4 campaigns; and (ii) 85 cruise measurements from the NASA ACCESS and ECLIF2/ND-MAX campaigns. Tables S1 86 to S5 compiles the fuel properties, engine operating conditions, as well as the measured and 87 estimated nvPM EIn from the five experimental campaigns. We note that the ECLIF/ND-MAX 6 88 campaign at cruise measured the nvPM EIn for different fuel types independently without fixing 89 the fuel mass flow rate, and therefore, we used data from the International Civil Aviation 90 Organization (ICAO) Aircraft Emissions Databank (EDB) 8 and the methodology of Teoh et 91 al. 9 to estimate the nvPM EIn that would have been emitted under the "reference" Jet A-1 fuel, 92 and then scale the nvPM EIn from SAF using the two independent methodologies. The 93 coefficient of determination (R 2 ) and normalised mean bias (NMB) from both models are 94 presented in Figures S5 and S6. It shows that: (i) the performance of both models is comparable 95 when compared against ground measurements (R 2 = 0.84 and NMB = +28% for our extended 96 model, vs. R 2 = 0.78 and NMB = +6.8% for the ICAO CAEP/11); but (ii) our extended fuel 97 composition model outperforms the ICAO CAEP/11 approach when compared against cruise 98 measurements (R 2 = 0.83 and NMB = -3.2% for the extended model, vs. R 2 = 0.03 and NMB 99 = -13% for the ICAO CAEP/11). For the comparison with cruise measurements, we note that 100 the negative bias in the estimated ΔnvPM EI n from the ICAO CAEP/11 (NMB = -13%) is S7 consistent with the results shown in Figure S4a, where the estimated ΔnvPM EI n from the 102 ICAO CAEP/11 is between 0% and 8% smaller than those estimated from our extended fuel 103 composition correction model. This comparison provides supporting evidence that our 104 extended fuel composition correction model (Eq. 1 in the main text) is applicable for both 105 ground and cruise conditions. 106 107 Figure

S2
Fuel properties from different SAF blending ratios 145 The fuel properties of conventional "reference" fuels and SAF with different blending ratios 146 are compiled from the literature and presented in Table S6. We use the compiled dataset to 147 develop a linear relationship between the SAF blending ratio (pblend, in %) versus the fuel 148 hydrogen mass content (Hfuel, in %) and lower calorific value (LCV in J kg -1 ) ( Figure S7). On 149 this basis, 150 the following equations are used to approximate the SAF fuel hydrogen content (HSAF) and 151 LCV (LCVSAF) for different pblend, 152 where we assume Href = 13.8% and LCVref = 43.1 ×10 6 J kg -1 respectively for the "reference" 153 Jet A-1 fuel 7,10 ; and HSAF=100% = 15.3% and LCV SAF=100% = 44.2 ×10 6 J kg -1 respectively for a with HSAF, 156    Table S6. 166 We note the small deviations in the different fuel properties around the assumed linear 167 trendline, for example, the Hfuel measured from SAF's with the same blending ratio can differ 168 by up to ~0.5% ( Figure S7a). This phenomenon likely arises from: (i) variations in the Hfuel of 169 different conventional "reference" fuels (13.5 -14.1%, Table S6) that was used to blend with 170 the SAF; and (ii) the different technology pathways that was used to produce the SAF (i.e., 171 HEFA-SPK and FT-SPK). Therefore, the stated HSAF values for a given pblend in Table 1

189
In this study, we do not incorporate this step change in nvPM emissions for DAC and TAPS 190 engines and instead, linearly interpolate the nvPM EIn from the four data points provided by 191 the ICAO EDB because the transition point between the "rich-burn" and "lean-burn" phase is 192 not publicly available. However, we do not expect this assumption to change our simulation 193 results because the: (i) nvPM EIn inputs to the contrail cirrus prediction model (CoCiP) 16 is 194 constrained to a lower bound of 10 13 kg -1 to account for uncertainties and the potential 195 activation of ambient aerosols and organic volatile particles into contrail ice crystals (refer to 196 10 14 kg -1 in the simulation with fully synthetic SAF (SAF100), as shown in Figure S8. The 198 small number of flights with nvPM EIn < 10 14 kg -1 (~0.5%) can be attributed to the low usage 199 of aircraft types that are powered by the TAPS combustor (Boeing 737-MAX, 747-800, 787-S14 10 and the Airbus A320neo) over the North Atlantic (refer to Figures S6 and S7 of Teoh et 201 al. 9 ). Therefore, for all SAF simulations, the mean nvPM EIn are in the "soot-rich" regime 17 202 and exceeds 10 13 kg -1 by more than one order of magnitude (Table 1 in the main text). 203

S3.2 Contrail properties 204
The probability density functions in Figure S9 show the change in persistent contrail formation 205 and contrail energy forcing (EFcontrail) for all contrail-forming flights when SAF with different 206 blending ratios are used. Figure S10

S3.3 Comparison with existing studies 226
We compare the difference in contrail properties between the baseline scenario and SAF100 227 ( Table 2 in

240
All four studies assumed a constant nvPM EIn for all waypoints, while we account for 241 variations in nvPM EIn from different aircraft types (Section 2.2), and the mean reduction in 242 nvPM EIn from SAF (~52%) is a function of ̂ and HSAF (Section 2.3). We estimate a slightly 243 smaller increase in flight distance forming persistent contrails (+5.0%) when compared with However, the change in rice from Caiazzo et al. 18 (+58%) is around two times larger than our 247 study (+26%) because they assumed a larger reduction in the nvPM EIn (-75% vs. a mean of -248 52% in our study), where humidity in the contrail plume is distributed to fewer particles. Our 249 estimated change in τcontrail (-22%) and contrail cirrus cover (-41%) are within range of values 250 compiled from the comparison studies (τcontrail between -49% and -21%; and contrail cirrus 251 cover between -41% and -15%), and the large range between studies is due to differences in 252 the selected domain area and the assumed reduction in nvPM EIn (Table S7). 253 While the change in annual mean contrail cirrus net RF from our study (-44%) appears to be in 254 contrast with Caiazzo et al. 18 (-4% to +18%), these reported values likely represent the mean 255 contrail net RF', i.e., change in radiative flux per contrail area, instead of the annual mean 256 contrail cirrus net RF over a specific domain (refer to    Table S8 summarises the changes in contrail occurrence and annual contrail energy forcing 277 (EFcontrail) when the limited supply of SAF is blended at different ratios and targeted to flights