403
1 INTRODUCTION
Today modern merchant vessels have a huge cargo
carrying capacity. Very large bulk carriers (VLBC)
carryironoretoafullcapacityof400,000DWT(dead
weight tonnes), (Proposes). Ultra large crude oil
carriers (ULCC), have an oil carrying capacity of
5000,00DWT,(Ingleetal.,2010).Theworld’slargest
container vessel has a capacity of20,568twenty‐foot
equivalentunits(TEUs),(Kimetal.,2002).Moreover,
liquid natural gas (LNG) have a tonnage carrying
capacity of 128,900 DWT, with a cubic capacity of
266,000 cubic meters, (Pitblado et al., 2006). These
huge vessels are propelled by large capacity marine
diesel engines referred as a main engine. The main
engine can produce power to the tune of 100
MW
(megawatts).Tosafelytransportcargofromoneport
to another, it is important to ensure that the main
enginepropellingthegiantvesselsissafeandreliable.
Moreover,it isanongoingchallengeforcommercial
shippingoperatorsto dealwith issuesrelated tothe
protectionofenvironmentandlegalimplicationson
a
day to day basis (Hatzigrigoris et al., 2005).
Numerous marine accidents occurred due to the
failureofmainengines.InAugust2001,aHongKong
flagged, cellularcontainership of 44153 deadweight
tonnes“MaerskTacoma”hadanaccidentatseadue
to the failure of lubricating oil system of
a main
Data Analysis to Evaluate Reliability of a Main Engine
M.Anantharaman&R.Islam
A
ustralianMaritimeCollege,UniversityofTasmania,Launceston,Australia
F.Khan
M
emorialUniversityofNewfoundland,St.John’s,NL,Canada
V.Garaniya&B.Lewarn
A
ustralianMaritimeCollege,UniversityofTasmania,Launceston,Australia
ABSTRACT:Maritimetransportationistheessenceofinternationaleconomy.Today,aroundninetypercentof
worldtradehappensbymaritimetransportationvia50,000merchantships.Theseshipstransportvarioustypes
ofcargoandmannedbyoveramillionmarinersaroundtheworld.Majorityof
theseshipsarepropelledby
marine diesel engines, hereafter referred to as main engine, due to its reliability and fuel efficiency. Yet
numerousaccidentstakeplaceduetofailureofmainengineatsea,themaincauseofthisbeinginappropriate
maintenanceplan.Toarriveatanoptimalmaintenanceplan,
itisnecessarytoassessthereliabilityofthemain
engine.AtpresentthemainengineonboardvesselshaveaPlannedMaintenanceSystem(PMS),designedby
theshipmanagementcompanies,considering,adviseoftheenginemanufacturersand/orship’schiefengineers
andmasters.FollowingPMSamountstocarryingoutmaintenanceof
amainenginecomponentsatspecified
runninghours,withouttakingintoconsiderationtheassessmentofthehealthofthecomponent/sinquestion.
Furthermore,shippingcompanies havealimited technicalabilitytorecord the dataproperly and usethem
effectively. In this study, relevant data collected from various sources are analysed
to identify the most
appropriatefailuremodelrepresentingspecificcomponent.Thedatacollected,andmodeldevelopedwillbe
veryusefultoassessthereliabilityofthemarineenginesandtoplanthemaintenanceactivitieson‐boardthe
ship.Thiscouldleadtoadecreaseinthefailureofmarineengine,ultimately
contributingtothereductionof
accidentsintheshippingindustry.
http://www.transnav.eu
the International Journal
on Marine Navigation
and Safety of Sea Transportation
Volume 13
Number 2
June 2019
DOI:10.12716/1001.13.02.18
404
engine (ATSB, 2001). In July 2006 the Antigua and
Barbuda registered self‐discharging bulk carrier
Enterprise had an accident due to the failure of
lubricatingoilsystemofamainenginewhileen‐route
from Adelaide to Sydney, Australia. As a results of
thisfailuremainenginestoppedandcaused
blackout
oftheshipatsea.Therefore,theshiphadtobetowed
to the nearest port for repairs. The Australian
Transport Safety Bureau (ATSB) investigation
identified that maintenance planning for the main
lubricating oil pump was inadequate and engineers
did not follow the manufacturer’s instructions, even
though the pump
had failed previously. Also, the
execution of routine maintenance on the lubricating
oil filter was inadequate in that the spare filter was
notreadyforuse.Theshipboardproceduresdidnot
identify the error and the procedures for operating
andmonitoringthefilterwerealsoineffective(ATSB,
2006). The above
discussion clarifies that there is a
needtoaddressthepropermaintenanceplanningon‐
board ships. The operation on‐board ships at high
seasareverycomplexinnatureandisdependenton
the competencies of the personnel operating and
maintaining the machinery on‐board. Also, the
reliability of the system
components must be high.
This will ensure high reliability of the subsystems,
whichinturnwilldictatehighreliabilityofthemain
propulsion engine (Monieta, 2016). Assessment of
main engine failures done by various groups have
shown that the failures of pistons, piston rings,
cylinder liners and geometry of the
combustion
chamber have highest proportion every year
(Kamiski, 2017). Furthermore, research has been
conducted to study the behaviour of the system
components under various operational and
maintenance policies. Therefore, it is required to
evaluateanoptimummaintenanceplanwhichwould
ensurereliabilityofthemainengineandatthesame
time
economicallyviable(Baliwangietal.,2009).
Tomakebettermaintenanceplanitisrequiredto
assessthereliabilityofthemarineengines.However,
thereisalackofappropriatedataandmodelthatfits
the data. Therefore, collecting the relevant
appropriate data is very important. Currently there
arelimiteddata
availableintheliteraturehowever,it
is not collected in a structured way to develop
reliabilityassessmenttechnique.Tomeetthescientific
rigour and enable generalization of the data and its
interpretation, varioussourcesofdataandmodes of
feedback, such as interviews with experienced
seafarers’ on‐board, review of existing
documentation, and a direct questionnaire method,
canbeused. Itenablesdatacollection fromglobally
operatingrespondents.It widenstheapplicability of
the methodandhelps to generalize the data and its
interpretation.Itisalsoaneasy,effective,economical,
flexible, and fast technique for data collection and
developmentof
aconceptualframeworkandhasbeen
previously used by researchers, Szolnoki and
Hoffmann (2013), Attwood et al. (2006) Islam et al.
(2018),Islametal.(2017).Therefore,thisapproachis
adoptedinthisstudy.Thecollecteddataisanalysed
throughaseriesofstatisticaltechniquestocheckthe
diversity and generalization
of the data and its
interpretation. Moreover, the collected relevant data
fromallthedifferentsourcesusedtoidentifythemost
appropriate failure model representing specific
components of a marine engine. This model will be
veryusefultoshippingcompaniesforplanningtheir
maintenanceofthemainengine.
2 QUESTIONNAIRE
STRUCTURE
The generalstructureof the questionnaire discussed
inthissectiontobetterunderstandtheresponses.The
mainengineisassociatedwithseveralsubsystemsto
perform the task of propelling the huge merchant
vessels (ABS, 2004). The key subsystems of a main
enginearei)lubricatingoilsystemii)fueloil
system
iii)coolingwatersystemandiv)scavengeairsystem
(MollenhauerandTschöke,2010).Thereliabilityofa
marineengineisaproductofthereliabilityofallthe
subsystems(i.e.lubricatingoilsystem,fueloilsystem,
cooling water system and the scavenge air system).
Therefore, thequestionnaire structured in
this study
such a way that responses can be used to evaluate
reliabilityoftheeachsubsystemandfinallytoassess
thereliabilityofamainengine(Mokashietal.,2002).
Thequestionnaireisgiveninthetable1below.
Table1.Questionnairetoseekthefeedbackfromexperience
marineengineers
_______________________________________________
1Pleasewritethenameoftheengineandmodelnumber
youhaveworkedwith(e.g.MANB&W6SMC60)inthebox
below.
2PleaseprovideFailureRunningHours(FRH)forthe
followingcomponentinthetablesbelow.Forexample,if
thePlannedMaintenanceHours(PMH)is500andthe
component
fails100hrsbeforePMH,itmeans(FRH)forthe
componentis400.PleasenoteFRH<PMH.
_______________________________________________
_______________________________________________
Question1seeksaresponsetoidentifythetypeof
engineanditsmodel(e.g.MANB&W6SMC60).
Questions 2 seeks feedback to know the FRH for
each individual components of a main engine sub‐
systems.
405
3 SELECTIONOFTHERESPONDENTS
To complete the survey, a number of experienced
marine engineers were identified in the shipping
industry.Thepotentialrespondentselectedbasedon
thefollowing criteria:i)at least5‐10yearsofengine
maintenance experience on‐board ship, ii) has been
sailingas3
rd
engineer,2
nd
engineerorchiefengineer
for ships engine department. A SurveyMonkey link
was created to conduct the questionnaire survey.
Ethicsapproval wassought,asper theguidelinesof
the University of Tasmania. Therefore, a human
research ethics approval was obtained from the
University of Tasmania’s human research ethics
committee (Ethics Ref
No: H0014474). The
SurveyMonkey link was sent around the world by
email to a total of 200 experienced ship’s engineers.
The total responses received from the 101
respondents. In other words, the response rates are
50.5%.Responsestothesequestionswereanalysedto
qualify the subjectivity and uncertainty in the
responses.
Tostatisticallyvalidatetheaccuracyofthe
collected responses, the required sample size is
estimatedusingEquation1.
Requiredresponses
2
2
ZP1 P
n
e
(1)
(IslamandYu,2018,Islametal.,2018,Islametal.,
2017)
whereeisthemarginoferror(e=±0.10);Zisnormal
scalevaluecorrespondingto95%confidence. P
is
the level of satisfaction; it is considered to have the
medianvalueof0.50.Resultsoftherequiredsample
size demonstrate that it is necessary to have 96
responsesfromeachdepartmenttostatisticallyjustify
theaccuracyofthecollectedresponse.Theresponses
reported in this study are more than
the required
number of responses. This confirms the validity of
enough responses and assumption of normality
distributionofresponses.
4 STATISTICALANALYSISOFTHEDATA
Statistical analysis is the science of collecting,
examining, interpreting and presenting data to
determine the basic form, relationships, and trends.
Statisticalanalysisforresearchis
necessaryasitoffers
clarificationofseveralconcepts,theories,frameworks
and methods. Moreover, it helps in arriving at
conclusionsandprovidingthehypothesis.Therefore,
aftercollectingthedata,theFRHwascomputedand
statisticalanalysiswascarriedout.Aftercollectingthe
data,aboxplotofthedatasetwasdrawn
inorderto
eliminate the outliers. A box plot is a method for
individually and removed the outliers.The
frequency plot for one of the sub‐systems, viz. the
cooling water system provided in Figure 1 below
representingstatisticaldataonaplottovisualizekey
statistical measures. The box
plot drawn for all the
componentsofasub‐systems.
8000600040002000
25
20
15
10
5
0
Failure Running Hours (FRH)
Frequency
a)
2800024000200001600012000800040000
18
16
14
12
10
8
6
4
2
0
Failure Running Hours (FRH)
Frequency
b)
2500020000150001000050000
40
30
20
10
0
Failure Running Hours (FRH)
Frequency
c)
3000024000180001 200060000
20
15
10
5
0
Failure Running Hours (FR H)
Frequency
d)
2800024000200001600012000800040000
30
25
20
15
10
5
0
Failure Running Hours (FR H)
Frequency
e)
Figure1.FrequencyplotofCoolingWaterSystem:a)Fresh
WaterCooler,(b)CoolingWaterPump,(c)ExpansionTank,
(d) Fresh Water Heater, (e) Fresh Water Temperature
ControlValve.
ThenextpartofthedataanalysistoproduceaBox
plot,NormalplotandtheWeibullplotforthecooling
watersystemcomponents.The sample plots for two
components;coolingwaterpumpandthefreshwater
heaterareshowninFigures2and3respectively.
CoolingWaterPump
30000
25000
20000
15000
10000
5000
0
Failure Running Hours (FRH)
Boxplot of Cooling Water Pump
10000010000100010010
99.9
99
90
80
70
60
50
40
30
20
10
5
3
2
1
0.1
Shape 1.379
Scale 11106
N92
AD 0.500
P-Value 0.216
Failure Running Hours (FRH)
Percent
Cooling Water Pump
Weibull - 95%
400003000020000100000-10000-20000
99.9
99
95
90
80
70
60
50
40
30
20
10
5
1
0.1
Mean 10143
StDev 7449
N92
AD 2.772
P-Value <0.00 5
Failure Running Hours
Percent
Cooling Water Pump
Normal - 95%
Figure2. Box plot, Weibull plot and Normal plot for
Coolingwaterpump
40000
30000
20000
10000
0
Failure Running Hours (FRH)
Boxplot of Fresh Water Heater
10000010000100010010
99.9
99
90
80
70
60
50
40
30
20
10
5
3
2
1
0.1
Shape 1.358
Scale 11789
N92
AD 1 .678
P-Value <0.010
Failure Running Hours (FRH)
Percent
Weibu ll - 9 5% CI
Fresh Water Heater
400003000020000100000-10000-20000
99.9
99
95
90
80
70
60
50
40
30
20
10
5
1
0.1
Mean 1082 2
StDev 7890
N92
AD 2.500
P-Value <0.0 05
Failure Running Hours (FRH)
Percent
Normal - 95% CI
Fresh Water Heater
Figure3.Boxplot,WeibullplotandNormalplotforFresh
waterheater
406
Table2.Tabulationofgoodnessoffittestforcoolingwatersystem
__________________________________________________________________________________________________
CoolingWaterSystemDistributiontypeBestfit
NormalExponential WeibullGammaDistribution
P AD P AD P AD P AD
value value value value value value value value
__________________________________________________________________________________________________
FreshWaterCooler0.005 8.725 0.003 5.690 0.010 3.493 0.005 2.895 Gamma
CoolingWaterPump0.005 2.772 0.003 2.619 0.500 0.216 0.250 0.477 Weibull
ExpansionTank0.005 8.122 0.003 6.903 0.010 4.789 0.005 4.683 Gamma
FreshWaterHeater0.005 2.500 0.003 3.444 0.010 1.678 0.005 1.906 Weibull
FreshWaterTemperatureControl
Valve 0.005 2.437 0.003 3.714 0.010 1.680 0.005 1.841 Weibull
__________________________________________________________________________________________________
R=Exp(‐t/11975.4)^1.0705R=Exp(‐t/14630.5)^1.1264
(Aplha1.075,Betais11975.4)(Alpha1.1264,Betais4630.5)
Figure4.Weibullmodelforcoolingwaterpumpandfreshwaterheater
After drawing the probability plots for all the
components of the cooling water system, the results
weretabulatedasshowninTable2.
It is seen from the table above that 60%
components of the cooling water system fitted the
Weibull distribution and other 40% fitted Gamma
distribution. Accordingly, it is
suggested that the
Weibull model could be a better fit for the cooling
watersystem.
The data for the cooling water pump and fresh
water heater was used to develop the Reliability
modelbased onWeibulldistribution TheAlphaand
Beta values were estimated and the Reliability
equationfor each
ofthe component established. The
ReliabilityforeachofthecomponentagainsttheFRH
is shown in Figures 4 below. A Chief Engineer on
board the vessel can accordingly schedule his
maintenance activities to achieve a high level of
reliability.
5 CONCLUSIONS
This study provides useful data for the reliability
analysis
of a main propulsion engine. The collected
dataisuniqueinthisfieldofstudy.Thelargesetof
collected data enables generalization and the
processeddatawillhelptodevelopreliabilityanalysis
techniques. The subjectivity and variability of the
collecteddataareanalysed,anditisfoundtobe
less
than10%. Therefore, itprovidesahigherconfidence
inthedataanditsgeneralization.Theanalysisofthe
data collected through a structural survey
demonstrates significance of the main propulsion
engines’sub‐systems andits components during the
voyage. The collected relevant data from various
sources are used to identify
the most appropriate
failure model representing specific components of a
main propulsion engine. The results of this study
indicate,not all the data for a component of a main
propulsion engines’ sub‐system follow the same
behaviour. Based on the observations and present
data a Weibull model is suggested to
estimate
reliability of the cooling water system of the main
engine.
REFERENCES
ATSB2006.Independentinvestigationintothemainengine
failure on board the self discharging bulk carrier
Enterprise.
ATTWOOD,D.,KHAN,F.&VEITCH,B.2006.Offshoreoil
and gas occupational accidents—what is important?
Journal of loss preventioninthe processindustries,19,386‐
398.
BALIWANGI,L.,ARTANA,K.B.&
ISHIDA,K.Riskbased
multi‐objectivesimulationofshipmainenginesystems.
28th International Conference on Ocean, Offshore and
ArcticEngineering,OMAE2009,May 31,2009‐June 5,
2009, 2009 Honolulu, HI, United states. American
SocietyofMechanicalEngineers(ASME),181‐188.
HATZIGRIGORIS, S., DRITSAS, C., HELIOTIS, A.,
SPERTOS, A., MORFINOS, C.
& KRIEZIS, G. Day‐to‐
day challenges of a shipping company from the
technicaldepartmentʹspointofview‐Lessons learned.
1st International Symposium on Ship Operations,
Management and Economics 2005, 2005. Society of
NavalArchitectsandMarineEngineers,173‐198.
INGLE, R. B., SAXENA, B. K. & KHAN, M.
A. J. 2010. A
theoretical investigation of natural frequencies of
vibrationandnoiseduetoengineandpropellersystems
ofanUltraLargeCrudeCarrier(ULCC).Journal of Low
FrequencyNoise,VibrationandActiveControl,29,41‐53.
ISLAM,R.,ABBASSI,R.,GARANIYA,V.&KHAN,F.2017.
Development of
a human reliability assessment
techniqueforthemaintenanceproceduresofmarineand
offshore operations. Journal of Loss Prevention in the
ProcessIndustries,50,416‐428.
407
ISLAM,R.,KHAN,F.,ABBASSI,R.&GARANIYA,V.2018.
Humanerrorassessmentduringmaintenanceoperations
ofmarinesystems–Whataretheeffectiveenvironmental
factors?Safetyscience,107,85‐98.
ISLAM, R., ABBASSI, R., GARANIYA, V. & KHAN, F.
2017. Development of a human reliability assessment
techniqueforthemaintenanceprocedures
ofmarineand
offshore operations. Journal of Loss Prevention in the
ProcessIndustries,50,416‐428.
ISLAM,R.,KHAN,F.,ABBASSI,R.&GARANIYA,V.2018.
Humanerrorassessmentduringmaintenanceoperations
ofmarinesystems–Whataretheeffectiveenvironmental
factors?Safetyscience,107,85‐98.
ISLAM, R. & YU, H. 2018.
Human Factors in Marine and
Offshore Systems. In: FAISAL KHAN, R. A. (ed.)
MethodsinChemicalProcessSafety.Elsevier.
KAMISKI, W. Assessment and analysis of selected main
engines cylinder liners and piston rings wear process.
17th International Multidisciplinary Scientific
Geoconference, SGEM 2017, June 29, 2017‐July 5, 2017,
2017 Albena,
Bulgaria. International Multidisciplinary
ScientificGeoconference,97‐104.
KIM,S.‐E.,CHOI,S.‐H.&VEIKONHEIMO,T.Modeltests
on propulsion systems for ultra large container vessel.
The Twelfth International Offshore and Polar
Engineering Conference, 2002. International Society of
OffshoreandPolarEngineers.
MONIETA,J.2016.Shipfirescausedbyprimaryfailure
of
the fuel filter of the engine driving the generator.
Scientific Journals of The Maritime University of Szczecin,
Zeszyty Naukowe Akademii Morskiej w Szczecinie,119,36‐
41.
PITBLADO, R., BAIK, J. & RAGHUNATHAN, V. 2006.
LNG decision making approaches compared. Journal of
hazardousmaterials,130,148‐154.
PROPOSES,A.
C.BulkShippingConceptAimstoExpand
CapacityofShallow‐waterPorts.
SZOLNOKI, G. & HOFFMANN, D. 2013. Online, face‐to‐
face and telephone surveys—Comparing different
sampling methods in wine consumer research. Wine
EconomicsandPolicy,2,57‐66.
