Analysis of Fuel Consumption and Emissions from Ships

Analysis of fuel consumption and emissions from the ship
Statistical Data Collection and Interpretation
Student Name:
Instructor Name:
Course Number:
9th May 2019

ABSTRACT:
A model is developed to try and predict the fuel consumption of the sea going ships in Australia.
The main aim of this study was to apply statistical techniques to analyze the relationship between
the fuel consumption and the different emissions coming from the ship. Results showed that
there is a strong positive relationship between fuel and the emissions coming from the ship.
Regression analysis showed that three variables significantly predict the fuel consumption. The
three variables include; particle (PM10), carbon monoxide (CO) and carbon dioxide equivalent
(CO2e).
INTRODUCTION:
Exhaust gas emissions as well as particles coming from seagoing ships play an important role in
influencing the total emissions in the transport sector worldwide (Wang, et al., 2010; Ng, et al.,
2013). They influence the atmospheric chemical composition, territorial air quality, climate and
wellbeing. The key emissions coming from the ship include nitrogen oxides (NOx), carbon
monoxide (CO), carbon dioxide (CO2), particles (PM), sulfur dioxide (SO2) and volatile organic
compounds (VOC). Emissions of NOx and SO2 from the ship may likewise give rise to localized
ocean acidification (Jalkanen, et al., 2012)
Eryring et al. (2010) and Dalsoren et al. (2009) evaluated worldwide mortalities resulting from
the ship engine fume discharges by applying a worldwide atmosphere model to a geospatial
shipping emissions inventory to decide overall convergences of PM2.5 from Ocean Going
Vessels.
The PM2.5 fixations were utilized in cardiopulmonary and lung malignant growth fixation
hazard capacities and populace models to evaluate yearly untimely mortality from these

outflows. It was finished up that without extra controls on fuel sulfur content, around 87,000
unexpected losses would happen in 2012. For Australia, the number of untimely mortalities
because of ship outflows is assessed at around 90 to 300 for each annum (Winebrake, 2013).
Transportation offers a generally low ozone depleting substance choice for transport of
merchandise (Buhaug et al. (2009)). Despite the fact that delivery is a significant component of
the Australian economy, deliver motor fumes outflows in Australian waters have not been
thoroughly examined. There is restricted learning about the fumes outflows from ship motors and
boilers in seaside districts and ports in Australia and the impacts of these outflows on air quality
in adjacent urban locales. This issue is of developing centrality on account of the expanded
guideline of land based emanations, the restricted guideline of transportation outflows and
arranged increments in delivery movement. Seaside and in-port ship emanations, when advanced
over land may cause disintegration in air quality. Moreover, the beach front outflows that can be
therefore advanced over land are not commonly considered in Australian examinations,
regardless of them being a lot bigger than in-port emanations. The main purpose of this study
was to present statistical analysis on the relationship between the fuel consumption of the ships
and the emissions that come out from those ships. The research question that this study sought to
answer was on whether there is a significant relationship between the fuel consumption and the
emissions made by the ships.
METHODOLOGY:
An experimental research was performed where secondary data was gathered to determine the
statistical distributions related to the seagoing ships emissions. The sample data was collected for
about 34 ports. Some of the variables collected include; nitrogen oxides (NOx), carbon
monoxide (CO), carbon dioxide (CO2), particles (PM), sulfur dioxide (SO2) and volatile organic

compounds (VOC). Some of the statistical parameters used to analyze the data include; the
measures of central tendency (mean and median) as well measures of variance (range and
standard deviation). The distributions of the variables was also determined. An inferential
statistics involving Pearson correlation test was performed to investigate the relationship that
exists between the variables.
Regression analysis was also employed to try and predict the fuel consumption based on the
amount of emission particles from the ship. The regression model that this study sought to
estimate is given as follows;
y=β0 + β1 x1 + β2 x2 + β3 x3 + β4 x4 + β5 x5 + β6 x6 + β7 x7 +ε
Where,
y=fuel consumption , x1=sulphur dioxide ( S O2 ) , x2=carbon monoxide ( CO ) , x3 =nitrogen oxide ( N Ox ) , x4= partic
β0= intercept (constant) coefficient, β1 , ... , β7are the coefficents for the predcitor variables.
COLLECTED DATA:
Data was collected from the ships using Automatic Identification System (AIS). The collected
data was for the period 2010/11 and it was collected from various coastal parts in Australia
including but not limited to Melbourne, Portland, Bunbury, Albany, Broome, Port Alma, and
Darwin among many other ports within Australia. The collected data set is presented in table 1
below.
Table 1: Data
Fuel SO_2 CO NO_x PM_2.5 PM_10 CO_2e VOC
3266 144 12 144 15 17 10595 4
5632 293 24 314 32 34 18242 8
703 32 3 33 3 4 2282 1

1151 58 4 51 6 6 3734 2
1620 75 6 75 8 9 5258 2
15127 709 58 801 78 85 49021 21
62287 2869 262 3131 315 343 201753 87
8020 411 36 400 45 49 25954 12
24215 1088 103 1246 121 131 78397 36
40352 2113 164 1896 225 245 130725 60
12831 403 50 585 44 47 41465 13
3440 158 13 149 17 18 11156 5
1792 76 7 78 8 9 5815 2
9366 461 23 300 45 49 30482 9
7877 375 28 339 39 43 25560 10
1158 54 5 55 6 6 3754 2
2059 95 9 105 10 11 6671 3
19535 1022 62 698 103 112 63454 23
1645 72 6 77 8 8 5334 2
871 37 4 44 4 4 2821 1
76383 3940 277 4375 440 478 247618 99
1808 91 8 97 10 11 5855 3
506 26 2 28 3 3 1638 1
413 14 1 17 1 2 1341 0
445 22 2 18 2 2 1445 1
12499 555 52 588 60 65 40492 17
8812 442 34 401 47 51 28565 12
539 23 2 26 3 3 1744 1
6085 234 25 287 25 28 19716 8
3147 149 13 164 16 18 10198 4
2371 107 9 109 11 12 7688 3
2373 107 9 101 11 12 7698 3
60752 3045 192 3140 333 362 197155 75
38135 1798 127 1597 187 204 123790 46
DESCRIPTIVE STATISTICS
Table 2 below presents the summary statistics for the various variables used in the study. The
average fuel consumed by the ships was found to be 12,859.26 with a median of 3353. The huge
variation in the mean and median for the fuel variable points a skewed distribution. Further
considering the standard deviation for the fuel we can clearly see that the data is widely spread

out as the standard deviation is very large (SD = 19700.66), actually the standard deviation was
found to be larger than even the mean. The skewness value for the fuel confirmed that the
distribution of fuel is heavily skewed (positively skewed). Considering nitrogen oxide (NOx), we
can see that the average emissions of this chemical was 631.44 with a median value of 156.50.
Again we can see that there is a very large variation between the mean and the median values for
the NOx which points to a non-symmetrical distribution of the NOx. The standard deviation (SD
= 1038.78) is very large as compared to even the mean which clearly shows that the data is
widely spread out. The skewness value of 2.40 further confirms the fears that the distribution is
skewed. With this skewness value we can confirm that the distribution of NOx is heavily skewed
(positively skewed).
For the Sulphur dioxide (SO2), it can be seen that the average emissions was 620.53 with a
median value of 153.50. Again we can see that there is a very large variation between the mean
and the median values for the SO2 which points to a non-symmetrical distribution of the SO2.
The standard deviation (SD = 982.07) is very large as compared to even the mean which clearly
shows that the data is widely spread out. The skewness value of 2.18 further confirms the fears
that the distribution is skewed. With this skewness value we can confirm that the distribution of
SO2 is heavily skewed (positively skewed).
Considering particle (Pm10), we can see that the average emissions was 72.09 with a median
value of 18.00. There is a very huge difference between the mean and the median as can be seen.
The standard deviation (SD = 117.31) was also found to be very large as compared to the mean
which clearly shows that the data is widely spread out. The skewness value of 2.25 clearly shows
that the distribution for the emissions of PM10 is very much skewed to the right.