249
1 INTRODUCTION
AccordingtotheJapanTransportSafetyBoard(2015),
there are distinctions between marine accidents and
marine incidents. The term marine accident refers to
the event wherein there is damage to a ship or
facilities other than a ship, related to the ship
operation, or causing death or injury to people
concerned with the constructi
on, equipment or
operation of a ship. Furthermore, marine incident
refers to the situation wherein the ship experiences
lossofcontrolduetonavigationalequipmentfailure,
listing of the ship, and shorts in the elementsystem
forengineoperation.
Themarineindustryisagreatglobalmarket,and
itisoneofthemostcapital‐int
ensiveindustriesdueto
the tremendous cost of the developed equipment
used (Ashmawy, 2012). Therefore, to increase the
productivity of this industry, it is essential to apply
effectivesafetymeasures.However,marineaccidents
arestillcommon(EMSA,2015).Humanerrorisoneof
the largest causes of ma
rine accidents. Hence,
nowadays, studies are being conducted with the
objectiveofreducingtheprobabilityof human error
inthemarine industry.Inrecentyears, international
organizationsthatengagedinthemaritimeactivities,
particularly authorities, such as the International
MaritimeOrganization(IMO),theInternationalLabor
Organization, and Ship Classification Societies
(IACS), have shown greater concerns regarding
humanerror(Akyuz,etall,2016).
Human Reliabilit
y Assessment (HRA) is the
analyticaltoolusedtoassessthecauseofaccidents.In
general, the methodology of HRA consists of two
steps: the qualitative method and the quantitative
method.However,therearealsoHRAswhichconsist
ofmoretha
ntwosteps,orofonlyonestep.
The Development of Marine Accidents Human
Reliability Assessment Approach: HEART Methodology
and MOP Model
L.P.Bowo&W.Mutmainnah
KobeUniversity,GraduateSchoolofMaritimeScience,Kobe,Japan
M.Furusho
KobeUniversity,Kobe,Japan
ABSTRACT: Humans are one of the important factors in the assessment of accidents, particularly marine
accidents.Hence,studiesareconductedtoassessthecontributionofhumanfactorsinaccidents.Therearetwo
generations of Human Reliability Assessment (HRA) that have been developed. Those methodologies are
classifiedbythedifferencesofviewpointsofproblem‐solving,asthefirstgenerationandsecondgeneration.
The accident analysis can be determined using three techniques of analysis; sequential techniques,
epidemiological techniques and systemic techniques, where the ma
rine accidents are included in the
epidemiological technique. This study compares the Human Error Assessment and Reduction Technique
(HEART)methodology and the 4M Overturned Pyramid (MOP) model, which are applied to assess ma
rine
accidents.Furthermore,theMOPmodelcaneffectivelydescribetherelationshipsofotherfactorswhichaffect
theaccidents;whereas,theHEARTmethodologyisonlyfocusedonhumanfactors.
http://www.transnav.eu
the International Journal
on Marine Navigation
and Safety of Sea Transportation
Volume 11
Number 2
June 2017
DOI:10.12716/1001.11.02.06
250
Indonesiaasanarchipelagiccountryusesshipsas
theirmeansoftransportationtoconnect eachisland.
Moreover, the vision of Indonesia is to become a
globalmaritimeaxis.Incontrast,thequalityofsafety
atsea in Indonesia is still low, which has led to the
occurrence of several accidents
(Bowo, Furusho,
2016).
The objective of this research is to compare the
HRAs in the marine industry, and to find a proper
methodology that can be applied to the marine
industry,especiallywithrespecttoaccidents.
The remainder of this paper is organized as
follows. The second chapter is a
literature review
about HRA generations and the characteristics of
marine accidents. The third chapter presents an
explanation about Human Error Assessment and
ReductionTechnique(HEART)methodologyandthe
4M Overturned Pyramid (MOP) model. Results of
thispaperwillbepresentedinthefourthchapter;and
the discussion and conclusions will be
presented in
chapterfiveandchaptersix,respectively.
2 LITERATUREREVIEW
2.1 HRAGenerations
Since the 20
th
century, researchers have been
developing HRA. Essentially, HRA has three
functions,namelytheidentificationofhumanerrors,
thepredictionoftheirlikelihood,andthereductionof
their likelihood, if required (Kirwan, 1996). Those
HRA functions were developed to assess the
probabilityoferrorinnuclearpowerplants.
Hollnagel summarized HRA
development from
1975–2005,asshowninFigure1below.Inthe1980s,
the development of HRAs had the largest growth,
when compared with other years. Moreover, this
period represents the first generation of HRAs.
Furthermore, in the 1990s, there was also
developmentofHRAs,althoughnotassignificantas
in the
1980s. Moreover, this period represents the
launchoftheso‐calledsecond‐generation(Hollnagel,
2005).
Figure1.CumulatedNumberofHRAMethodspublication
(Hollnagel,2005).
However,as ofrecent, allindustrial sectors,such
astherailway,airplane,medical,andmarinesectors,
use HRA to identify the errors that cause accidents
andincidents.Therefore,thedevelopmentofHRAsis
still ongoing. Owing to the large number of HRA
methodologies, the methodologies are classified by
the differences of
viewpoints of problem‐solving, as
firstgenerationandsecondgeneration.
2.1.1 FirstGeneration
ThefirstgenerationofHRAwasfirstdevelopedin
the 1980s. These HRAs were developed to help risk
assessors predict and calculate the likelihood of
human error. Furthermore, the first‐generation
methods focus on the skill and
rule base level of
human action, and are often criticized for failing to
consider aspects such as the impact of context,
organizationalfactors,anderrorsofcommission(Bell
& Holroyd, 2009). The methodologies which are
includedinthefirstgenerationareasfollows:THERP
(Techniquefor Human Error Rate Prediction), ASEP
(Accident Sequence Evaluation Program), HEART
(HumanErrorAssessmentandReductionTechnique),
andSPAR‐H(SimplifiedPlantAnalysisRiskHuman
ReliabilityAssessment).
2.1.2 SecondGeneration
More modern methods, the second generation of
HRAiscarefullyconsideredandmodelstheinfluence
ofcontextontheerror.Moreover,itutilizesfindings
and insights
from the then developed cognitive
movement (Boring, 2012). The development of this
secondgenerationbeganinthe1990s,andisgoingto
bedevelopedevenfurther.ATHEANA(ATechnique
for Human Event Analysis) and CREAM (Cognitive
Reliability and Error Analysis Method) are included
inthesecondgeneration.
2.2 MarineAccidentsCharacteristics
In all sectors, the development of accident analysis
canbedeterminedusingthreetechniquesofanalysis:
sequential techniques, epidemiological techniques,
and systemic techniques (Underwood & Waterson,
2013).
There are several differences between the three
techniques,asdescribedbelow.
2.2.1 SequentialTechniques
This is a simple, linear cause‐and‐effect model,
where accidents are modeled as a series of falling
dominos,which occurinaspecificand recognizable
order(U.S.DepartmentofEnergy,2012).Thismethod
describes the events leading up to accidents, using
physicalcomponentfailuresortheactionsofhumans
(Leveson,2011).
2.2.2 EpidemiologicalTechniques
An epidemiological technique can also
be
recognized as the latent failure model. With this
technique, accidents are seen as a combination of
unsafe acts (active failures) and unsafe conditions
(latentconditions)(U.S.DepartmentofEnergy,2012).
2.2.3 SystemicTechniques
A systemic technique describes losses as the
unexpected behavior of a system. In other words,
251
accidents arenot created by a combination of active
failuresandlatentconditions,butarerathertheresult
of humans and technology operating in ways that
seemrational(Underwood&Waterson,2013).
These are some HRA methods that are already
classifiedasthethree‐analysistechnique,asshownin
Table
1below.
Table1. Methods for Sequential Techniques,
Epidemiological Techniques and Systemic Techniques
(Underwood&Waterson,2013,U.S.DepartmentofEnergy,
2012)
_______________________________________________
TechniquesHRAMethods
_______________________________________________
SequentialFaultTreeAnalysis(FTA),EvenTree
Analysis(ETA),CriticalPathModel.
Epidemiological SwissCheeseModel,HumanFactors
Analysis&ClassificationSystem
(HFACS).
SystemicSystemsTheoreticAnalysisModeland
Processmodel(STAMP),Functional
ResonanceAnalysisMethod(FRAM),
Accimap.
_______________________________________________
BasedonFigure2,marineaccidentsarelocatedin
the 1
st
quadrant, which has high manageability and
tight coupling. The manageability itself means that
the principles of the function of the system are
known, and that system descriptions are simple,
having few details. Further, the system does not
change while it is being described (Underwood &
Waterson,2013).
Moreover, a tight
coupling system can be
described as having process sequences that are
invariant. In addition, the substitution of supplies,
equipment,and personnel is limitedand anticipated
inthedesignastightlycoupledsystemsaredifficult
tocontrol(Underwood&Waterson,2013).
Figure2. Accident Model Categorization (adapted from
(Underwood&Waterson,2013))
According to the accident model categorization,
thereisananalysistecniquewhichissuitableforeach
quadrant. In the first quadrant, the epidemiological
technique is suitable. Moreover, the systemic
technique is sutable for application to the second
quadrant, whereas the sequential technique works
wellwhenappliedinQuadrant4.
3
METHODOLOGY
3.1 HumanErrorAssessmentandReductionTechnique
(HEART)Methodology
Human Error Assessment and Reduction Technique
(HEART) methodology was first developed by
Williams (1986). This technique is based on the
human factors literature; it uses a set of basic error
probabilities,modifiedbytheassessor,bystructured
Performance Shaping Factors (PSF)
considerations
(Kirwan,1996).
The HEART methodology generally consists of
two assessment process steps. The first step is the
qualitativeprocess, wherein the assessor has tofind
the general task of the accidents, which consists of
eight points of the general task. After finding the
general task, it then breaks it
down to smaller parts
called Error‐producing conditions (EPCs).
Furthermore, EPCs represent unsafe acts of the
seafarers,whichleadtoaccidents.Thesecondstepof
thismethodologyisthequantitativemethod,wherein
Human Error Probability (HEP) is calculated. To
obtaintheHEPusingtheHEARTmethodology,first
obtain the nominal human
unreliability (NHU),
whichbelongstothegenerictask.WhentheEPCsare
determined,themultiplicationnumberofeachEPCis
obtained.Thereafter,theassessedimpactvalue(AIV)
canbecalculatedusingthefollowingformula:
11AIV EPC Multiplier APE
(1)
The assessed proportion effect (APE) is highly
judgmental,and no guidance isgiven in thecurrent
HEARTdocumentation(Kirwan,1996).Theresultof
thiscalculationwillbeusedtocalculate theHEP,in
order to determine the overall probability of failure
foreachcase.Theformulaisasfollows:
12
...
n
HEP NHU AIV AIV AIV
(2)
ThefinalresultoftheHEPcalculationisbetween0
and 1; where if the result is more than 1, it can be
assumed that the accident was definitely caused by
humanfactors(Bowo&Furusho,2016).
If the HEP results of the assessed accidents are
between 0 and
1, there is a probability that the
accidentsareinfluencedbyotherfactors.
3.2 4MOverturnedPyramid(MOP)Model
TheMOPmodelisanewHRAmethodology,which
wasdevelopedbyMutmainnahandFurushoin2014.
ThismodelisthedevelopmentoftheIMmodel.The
basic concept of IM model
is the individual (self)
252
centered properties, and the relationships between
others factors: man, machinery, media, and
management(Furusho,2002).
Currently, the MOP model has already been
developedtoassessthequalitativedataofaccidents.
Thefirststepistoobserveandbreakdowntheunsafe
acts,basedontheaccidentdata;namelyascausative
factors(CF).Inthisstep,itbreaksthemdowninto4
factors: man factors, machine factors, media factors,
and management factors. Afterwards, each CF is
relatedtoanotherCFwhichinfluencestheaccidents.
This relation is called Line Relation (LR). The basic
idea of LR is the relationship between the
accidents
occurred.
Figure3.The4MOverturnedPyramid(MOP)Model
(Mutmainnah&Furusho,2014).
ThequantitativeprocessoftheMOPmodelisthe
calculation of the probability percentage from each
factor, compared with the summation of all the
factors. Thereafter, the percentage contribution of
eachfactortotheaccidentsisobtained.
Inthisresearch,thefireandexplosionaccidentsin
Indonesiafrom2008–2013wereassessed.
Thedataof
the accidents in Indonesia were obtained from
IndonesiaNationalTransportationSafetyCommittee.
Subsequently, these data were assessed using the
HEARTmethodologyandMOPmodel.
4 RESULTS
The results show the comparison of EPCs using the
HEART methodology, and CFs using the MOP
model,andtheHEP.
4.1
Unsafeacts
Unsafe acts are the actions which lead to accidents.
ForeachHRAmethod,thenames ofunsafeacts are
different.Thisstepconsistsofthequalitativemethod
of the Human Reliability Assessment (HRA). In
HEART methodology, unsafe acts are known as
EPCs,andintheMOPmodelitis
knownasCFs.
Thereis atotalof38EPCs and 13 typesofEPCs
discoveredusing HEART methodologytoassess the
fire and explosion accidents in Indonesia, wherein
operatorinexperienceisthelargestcauseofaccidents.
Moreover,byusingtheMOPmodel,thereare69
CFs obtained from this
accident; where man factor
hasatotalof18points,machinefactorhasatotalof
35points,mediahasonlyonepoint,andmanagement
hasatotalof15points.Therefore,themachinefactor
isthemaincauseoftheaccidents,becauseithas the
largest number of points, when
compared with the
otherfactors. Theseresults are shown inTable 2 for
HEART methodology and Table 3 for the MOP
model,respectively.
Table2. Error Producing Conditions (EPC) by HEART
Methodology
_______________________________________________
ErrorProducingConditions(EPC)Total
_______________________________________________
Operatorinexperience5
Poorenvironment4
Spatialandfunctionalincompatibility3
Performanceambiguity3
Impoverishedinformation3
InadequateChecking3
Unreliableinstruments3
Unclearallocationoffunction3
Lowmorale3
Nodiversity2
Inconsistencyofdisplays2
Taskpacing2
Unfamiliarity1
_______________________________________________
Table3.CausativeFactors(CF)byMOPmodel
_______________________________________________
4MFactors CausativeFactors(CF)Total
_______________________________________________
ManIrresponsiblecrewor6
passengers
Slipshodworkmanship5
Incapabilityofseafarer4
Lackofutilizingequipment3
MachineEquipmentfailure11
Improperutilizationequipment 10
Damageshipconstruction5
Insufficientlayout4
Flammablematerialexistence 3
Equipmentoverload2
MediaWind1
Management Poorcargo
management5
Poormanagementofpersonnel 2
onboard
Poorcommunication2
PoorapplicationofSMS*2
Poormanagementofmaintenance 1
Poormanagementofberthing 1
schedule
Lackofsomenavigationand 1
safetyequipment
Poormanagementofmonitoring 1
andsupervisingfromcompany
orport
_______________________________________________
*SafetyManagementSystem(SMS)
4.2 HumanErrorProbability(HEP)
TheHEPforeachmethodologyisshowninthefigure
below. From 12 fire and explosion cases that were
already assessed using HEART methodology, the
finalresultoftheHEPs aremostlybelow0.5, which
indicatestheinvolvementofhumansasthecause of
accidents where
distract by other factors. Moreover,
there is only one case that is genuinely because of
human factors. The average result for the HEART
253
methodology HEP reveals that 23% of the accidents
werecausedbyhumanfactors.
Figure4. Human Error Assessment and Reduction
Technique(HEART)MethodologyHumanErrorProbability
(HEP)
Moreover, the MOP model has divided the
probability for every factor. Machine factors has the
biggestpercentage,about51%;thenfollowedbyman
(itcanbeassumedashumanerrorprobability),26%;
management, 22%; and media, 1%. Figures 4 and 5
showtheresultof HEPusing HEARTmethodology,
and
errorprobabilityusingtheMOPmodel.
Figure5.MOPmodelfactorpercentages
5 DISCUSSION
Theresults show that,there are similarities between
theEPCsfromtheHEARTmethodology,andtheCFs
from the MOP model, which are obtained from the
fireandexplosionaccidentsinIndonesia.IntheEPC
ofthe HEART,thereis operatorinexperience, and this
EPC is a rather
general description of inexperience;
whereas in the CF of the MOP model, the
inexperience of seafarers is shown as incapability of
seafarer and lack of utilizing equipment. Moreover, the
CF of Slipshod workmanship is similar to the EPC of
performance ambiguity, impoverished information,
inadequate checking and task pacing. Furthermore,
the
irresponsiblecreworpassengersinCFcanbeequatedto
thelowmoraleinEPC.
AsforthecommunicationproblemintheHEART
methodology, it is included in the impoverished
information and spatial and functional
incompatibility; whereas in the MOP model, they
separate the communication problem from the man
factor,andincludeitinthemanagementfactor.This
isbecauseaccordingtotheMOPmodel,thedefinition
of management is all elements that can control the
system and/or people, including communication
(Mutmainnah & Furusho, 2014). There are unreliable
instruments in the EPC that was already obtained in
theaccidentassessment
usingHEARTmethodology.
Moreover, it is related to the machinery problem in
theMOPmodel.
In HEART methodology, there are 38 EPCs that
were already established by Williams (1986). The
determination of EPC to utilize is based on the
assessorjudgement(Kirwan,1996).Thedevelopment
ofEPCswasachievedby
Akyuzetal.(Akyuz,Celik,
& Cebi, 2016), to generate EPC values for marine
transportation, by obtaining the new multiplication
numberoftheEPCinordertobespecifiedintermsof
shipoperationalmanagement.
Further,intheMOPmodel,theCFshavenotbeen
established yet, but there are several
CFs which
frequently appear during the assessment
(Mutmainnah & Furusho, 2016, Mutmainnah &
Furusho,2016).
ThefinalresultoftheHEPcalculationofHEART
methodologyandtheMOPmodelwerequitesimilar;
the average HEP of 12 cases that had been assessed
forHEART methodology was23%, and for the
man
probabilityitselfintheMOPmodel,theaverageHEP
was26%.ThisshowsthattheMOPmodelasthenew
methodology of HRA is been appropriate for use in
the assessment of accidents, such as the developed
methodology.
In the case of marine accidents, marine accidents
are included in the
category of epidemiological
techniques, having high manageability and tight
coupling.TheMOPmodelisasuitablemethodology
tobe applied in the assessment of marine accidents,
because the relationships between man and other
factors are described and detailed. Furthermore, the
MOP model is firstly proposed to be applied in the
marine industry (Mutmainnah & Furusho, 2014).
However,itispossibletodeveloptheMOPmodelto
assesstheaccidentsinothersectorsofindustries.
In addition, HEART methodology was firstly
proposedtosolvetheprobleminthenuclearindustry
(Williams,1986), which has a different category and
characteristic from marine accidents.
However, the
HEART methodology has been successfully applied
in many sectors of industries, such as the railway,
aviation,andoffshoreindustries(Deacon,etal.,2013),
by the regeneration of EPCs and the generic task
(Akyuz,etal.,2016).
6 CONCLUSIONS
Byconductingthisresearch,certainconclusionshave
beenarrivedat,
asfollows:
254
1 The4MOverturnedPyramid(MOP)modelcanbe
applied to assess the human reliability in
accidents.
2 In the case of marine accidents, the MOP model
can describe all the relationships between factors
which affect the accidents; whereas, HEART
methodologyisonlyfocusedonthehumanfactors
thataffect
theaccidents.
3 However,thecausativefactorsoftheMOPmodel
have not been established, and it is necessary to
generate the HEART methodology in order to be
suitableforcertainindustries.
REFERENCES
Akyuz, E., Celik, M., & Cebi, S. (2016). A phase of
comprehensive research to determine marine‐specific
EPC values in human error assesment and reduction
technique.SafetyScience,pp.63–75.
Ashmawy, M. E. (2012). The Maritime Industry and the
HumanElement Phenomenon. The 13th Annual General
AssemblyoftheIAMU,
pp.277–288.
Bell, J., & Holroyd, J. (2009). Review of human reliability
assessmentmethods.Derbyshire:HSEBooks.
Boring, R. L. (2012). Fifty Years of THERP and Human
ReliabilityAnalysis.Idaho:IdahoNationalLibrary.
Bowo, L. P., & Furusho, M. (2016). Human Error and
Reduction Technique for Marine Accidents Analysis:
The
Case of Ship Grounding. Asia Navigational
Conference.Yeosu.pp.193–201.
Bowo, L. P., & Furusho, M. (2016). Human Error and
ReductionTechniqueforreducingthenumberofmarine
accidentsinIndonesia.SENTA2016.Surabaya.
Deacon,T.,Amyotte,P.,Khan,F.,&S.MacKinnon.(2013).A
framework for human error analysis of
offshore
evacuatios.SafetyScience51,pp.319–327.
Dempsey,P.G.(2012).AccidentandIncidentinvestigation.
In G. Salvendy, Handbook of Human Factors and
Ergonomics. New Jersey: John Willey & Sons, Inc. pp.
1085–1091.
EMSA. (2015). Casualty with a ship. Annual Overview of
MarineCasualtiesandIncidents,p.31.
Furusho,
M. (2002). IM‐Model for ship safety. Inaugural
GeneralAssemblyProceeding,Turkey.pp.26–31.
Hollnagel, E. (2005). Human Reliability Assessment in
context.NuclearEngineeringandTechnology,pp.159–166.
JapanTransport Safety Board. (2015).Marine accident and
incident investigations. In Japan Safety Board Annual
Report2015.Japan.p.95.
Kirwan,
B.(1996).ThevalidationofthreeHumanReliablity
Quantificationtechniques‐THERP,HEART,andJHEDI:
Part 1‐technique descriptions and validation issues.
AppliedErgonomics,pp.359–373.
Leveson,N.(2011).Engineeringasaferworld:Systemsthinking
appliedtosafety.London:The MITPress.
Mutmainnah,W.,&Furusho,M.(2014).The4MOverturned
Pyramid (MOP) Model in Maritime Traffic System fo
SafetyatSea.Kobe.
Mutmainnah, W., & Furusho, M. (2016). 4M overturned
pyramid (MOP) Model Utilization: Case Studies on
Collision in Indonesian and Japanese Maritime Traffic
Systems (MTS). Transnav the International Journal on
Maritime Navigation and safety of sea transportation, pp.
257–264.
Mutmainnah,
W.,&Furusho,M. (2016).TheImproperlook‐
outthatleadstoshipcollisionsinJapan.AsiaNavigation
Conference2016,Yeosu.pp.443–449.
U.S. Department of Energy. (2012). Accident Models‐A
basic understanding. In Accident and Operational Safety
Analysis, Volume I: Accident Analysis Techniques.
Washington,DC.pp.1‐2‐1
‐4.
Underwood, P., & Waterson, P. (2013). Accident analysis
models and methods: guidance for safety professionals.
Loughborough:LoughboroughUniversity.
William, J. (1986). HEART‐a proposed method for
assessing and reducing human error. AEA Technology.
Culcheth:Warrington..
