_ 
zZ 
= 
Rai 
Measurements 
CO, —.— SO, 
Lg 
o 
Q 
5) 
168 
169 
470 
471 
472 
473 
474 
475 
476 
477 
478 
479 
480 
481 
482 
483 
484 
485 
486 
487 
488 
489 
490 
491 
492 
493 
494 
495 
496 
497 
A198 
OR za 
ee ee 
Time (s) 
) 
CC 
u 
Figure 9: Illustration of CO, (yellow line) and SO, (grey line) time-series from measurements. 
In case of Ship 1, the measured CO peak seems to be narrower in duration compared to 
the modelled one by approximately 50% (Figure 10a) but its height appears 2.3 times higher. The 
measured peak of Ship 2 appears rather distorted, as it presents a plateau at around 6.25 ppm, 
which decreases to 4.9 ppm in timestamps 30 s, 40 s and 45 s (Figure 10b) and does not show a 
typical increase and decrease as would be expected from a vessel approaching and then departing 
the MS. The modelled peak for Ship 2 appears to fix the weakness of measurement, although the 
predicted peak duration is 10 s shorter than the measurements. In the case of Ship 3, CFD 
modelling seems to overestimate concentrations, while the duration with the measured peak is 
identical (Figure 10c). Considering Ship 4, the measured peak is 10 s larger than the modelled one 
(Figure 10d) while the measured concentrations are lower than what the model predicts. Finally, 
the measurement of Ship 5 indicates two peaks, the second one being caused by the interference 
of a secondary source (Figure 10e). This example indicates that modelling can be useful to identify 
interferences of background sources. The detection duration for Ship 5 between measurement and 
modelling seems to be identical. 
The differences between the measurements and the modelling time-series may be due to 
several modelling uncertainties related to the ambient conditions, the ship technical characteristics 
and operation, etc. but also the measurements as such. Uncertainties related to the ambient 
conditions, include the random turbulence phenomena and temperature variations in the 
atmosphere that can hardly be measured and then modelled. With respect to vessel operation, 
estimates related to exhaust gas composition influence the results of CFD since the funnel 
concentrations calculation, in the current study, was performed using the combustion equation. 
assuming X = 2. The assumptions made for the FSC calculation, in parallel to the 30% uncertainty 
due to measurements, add further ambiguities in the exhaust concentration estimates. An additional 
factor of uncertainty is introduced by the temperature and the X hypotheses which are linked to the 
exhaust gas velocity estimation. Exhaust gas velocity effects on plume rise, influence the plume 
dispersion and the downwash phenomena which are, also, affected by the vessel’s structure and 
the relative angle between the ship and the wind speed vectors (Badeke et al., 2021; Dobrucali & 
Srgin, 2019; Li et al., 2022; Syms, 2004). Measurement uncertainty is related to the instrument 
specifications, removal of background concentrations and the influence of other sources. The 
7: