A Taxonomy of Performance Influencing Factors for Human Reliability Analysis of Emergency Tasks(1)
-
Upload
ali-abbasov -
Category
Documents
-
view
221 -
download
0
description
Transcript of A Taxonomy of Performance Influencing Factors for Human Reliability Analysis of Emergency Tasks(1)
-
Journal of Loss Prevention in the Process Industries 16 (2003) 479495www.elsevier.com/locate/jlp
A taxonomy of performance influencing factors for humanreliability analysis of emergency tasks
Jae W. Kim, Wondea JungIntegrated Safety Assessment Division, Korea Atomic Energy Research Institute, P.O. Box 105, Yuseong, Daejeon, 305-600, South Korea
Abstract
This paper introduces the process for, and the result of, the selection of performance influencing factors (PIFs) for the use inhuman reliability analysis (HRA) of emergency tasks in nuclear power plants. The approach taken in this study largely consists ofthree steps. First, a full-set PIF system is constructed from the collection and review of existing PIF taxonomies. Secondly, PIFcandidates are selected from the full-set PIF system, considering the major characteristics of emergency situations and the basiccriteria of PIF for use in HRA. Finally, a set of PIFs is established by structuring representative PIFs and their detailed subitemsfrom the candidates. As a result, a set of PIFs comprised of the 11 representative PIFs and 39 subitems was developed. 2003 Elsevier Ltd. All rights reserved.
Keywords: Performance influencing factors; Performance shaping factors; Human reliability analysis; Human error analysis; Emergency operation;Accident management
1. Introduction
Computerized automation has been adopted in largeparts of modern industrial high-risk and complex sys-tems such as nuclear power plants, aircraft and chemicalplants. However, humans still play important roles invarious parts of the design, maintenance, operation andsupervision of such systems. All human activities perfor-med in those parts are influenced by given specific work-ing conditions or task situations, so-called context,which is comprised of the MTO (man, technology andorganization) triad (Dougherty, 1993; Hollnagel, 1998).
In human error analysis (HEA) (Kirwan, 1992a,1992b) or human reliability analysis (HRA) in safetyassessment, such conditions that influence human per-formance have been represented via several context fac-tors. These context factors are referred to by differentterms according to method: PSF (performance shapingfactors), PIF (performance influencing factors), IF(influencing factors), PAF (performance affectingfactors), EPC (error producing conditions), CPC(common performance conditions), and so on. The PSFs
Corresponding author. Tel.:+82-42-868-8886; fax:+82-42-868-8374.
E-mail address: [email protected] (J.W. Kim).
0950-4230/$ - see front matter 2003 Elsevier Ltd. All rights reserved.doi:10.1016/S0950-4230(03)00075-5
or PIFs are used as causes or contributors to unsafe,human actions in event analysis (Paradies, Unger,Haas & Terranova, 1993; Nakanishi et al., 2002), andalso give a basis for assessing human factors in safetyassessment (Hollnagel, 1998; Embrey, 1984).
HRA has been performed as part of the probabilisticsafety assessment (PSA) of large-scale systems, such asnuclear power plants. PSA is an approach that developsall the possible accident scenarios and evaluates theoverall safety of a system probabilistically using theevent tree (ET) and fault tree (FT) techniques. The acci-dent scenarios are composed of two failure components,i.e. human failure events (HFEs) and hardware(system/component) failure events. HRA takes part inestimating the probability of those HFEs. There havebeen various approaches for evaluating humanreliability. In general, those approaches can be classifiedinto two categories, i.e. those using timereliability cor-relation (Hannaman et al., 1984; Moieni, Spurgin &Singh, 1994; Kim and Ha, 1997) and those manipulatingPIFs (Embrey, 1984; Phillips et al., 1990; Gertman etal., 1992). For the methods using PIFs, some of themuse a set of PIFs in adjusting the basic HEP (humanerror probability) such as THERP (Swain and Guttmann,1983), HEART (Williams, 1988), CREAM (Hollnagel,1998), and others in producing HEP by rating and inte-grating PIFs such as SLIM (Embrey, 1984), STAHR
-
480 J.W. Kim, W. Jung / Journal of Loss Prevention in the Process Industries 16 (2003) 479495
(Phillips, Humphreys, Embrey & Selby, 1990), etc. Onthe other hand, recently developed methods such asCREAM, INCORECT (Kontogiannis, 1997) andATHEANA (US NRC, 2000) use PIFs in the qualitativeanalysis/assessment of overall working conditions orerror-forcing context, as well as in the quantificationof HEP.
Generally, PIF taxonomies are so developed as to besuitable for a specific purpose and application area. HRAmethodologies also have their own PIF taxonomies, orthe PIF sets considered for HRA are somewhat differentdepending on the developer or the analyst. This maycause some important problems. The first one is of thetrust between the HEP values calculated from differentmethodologies that use a different set of PIFs. Also,comparison between different HRA results would bemeaningless, and, in order to obtain error reduction mea-sures, different PIFs would be dealt with according tothe used method. The second problem may be applicablefor some methods with a limited number of PIFs. Theuse of a limited number of PIFs might not only causeanalysts to omit important error reduction measures, butalso make the contribution of human error to the overallsafety of a system to be assessed lower than reality. Thethird problem is that the definition and the specific itemsto be assessed for each PIF are deficient or different bymethod, which can cause an inconsistent assessment ofthe individual PIF between assessors and, accordingly,produce different HRA results.
In this study, we suggest a new PIF taxonomy forhuman error/reliability analysis of emergency and acci-dent management tasks in nuclear power plants, withconsideration of the above-described problems of theexisting PIF taxonomies. The approach taken in thestudy to achieve the goal is summarized as follows.
Firstly, approximately 220 PIFs are collected from theexisting PIF taxonomies and other literature, and thosefactors are collated into a new set of PIFs. Among them,HRA PIFs are examined for the characteristics and trendof selection and usage, and the insights from the examin-ation are considered in the development of the new tax-onomy.
Second, the principal context under which humanoperators respond to emergency and severe accident situ-ations is analyzed, and PIFs relevant to such situationalcharacteristics are selected from the full-set of PIFs col-lated in the first step.
Third, the criteria of the PIF sets to be adequate forHRA are described, and, on the basis of these criteria,the candidate PIFs selected in the second step arerescreened out and structured in a two-layer hierarchy,i.e. the representative PIFs and their subitems.
The paper is structured as follows according to theabove-described approach. Section 2 includes a briefreview of the existing PIF taxonomies, the full-set ofPIFs collated, and the characteristics and trend of the
HRA PIFs. In Section 3, the principal context affectinghuman reliability during accident management and theselection of PIFs based on the context is described. Sec-tion 4 provides the final set of PIFs structured based onthe criteria of a PIF sets for HRA. Finally, Section 5concludes the study with a summary and remarks.
2. Collection and analysis of the existing PIFtaxonomies
2.1. Review of the existing PIF taxonomies
Two types of PIF taxonomies were collected in thestudy: one is composed of the detailed set of PIF, whichwas mainly developed for human error/factor analysis,and the other is the PIF set for the use in HRA. The firstone is referred to as the full set PIF taxonomy. Seventaxonomies among the detailed PIF sets and 11 taxo-nomies in HRA methodologies have been reviewed inthis study. HRA taxonomies are categorized accordingto the type of use as below. A brief description on thereviewed taxonomies is provided in Table 1. The list ofHRA PIFs is shown in Table 2.
The taxonomies of full-set of PIFs
CSNI taxonomy (Rasmussen, 1981) THERP (Swain and Guttman, 1983) HEART (Williams, 1988) PHECA (Whalley, 1987) PSF taxonomy (Bellamy, 1991) Influencing factors (Gerdes, 1997) K-HPES (KEPRI, 1998).
The taxonomies for use in HRA
Quantification of HEP: SLIM (Embrey, 1984), PLG-SLIM (Chu et al., 1994), INTENT (Gertman et al.,1992), STAHR (Phillips, Humphreys, Embrey &Selby, 1990), and HRMS (Kirwan, 1997)
Analysis of errors of commission: Macwans PIF tax-onomy for errors of commission (Macwan & Mosleh,1994), Julius PIF taxonomy for errors of commission(Julius, Jorgenson, Parry & Mosleh, 1995), andATHEANA (US NRC, 2000)
Overall context assessment and error analysis: HRMS,CREAM (Hollnagel, 1998; Hollnagel, Kaarstad &Lee, 1999), and INCORECT (Kontogiannis, 1997)
Database for HRA: Taylor-Adams PSF taxonomy forCORE-DATA (Taylor-Adams, 1995), and RogersPSF taxonomy for CORE-DATA (Gibson et al.,1998).
-
481J.W. Kim, W. Jung / Journal of Loss Prevention in the Process Industries 16 (2003) 479495T
able
1Su
mm
ary
ofth
eex
istin
gPS
F/PI
Fta
xono
mie
s
Met
hodo
logy
Des
crip
tion
PIF
taxo
nom
ies
for
HE
AC
SNI
taxo
nom
y(R
asm
usse
net
al.,
The
CSN
Ita
xono
my
was
deve
lope
dfo
rre
port
ing
inci
dent
san
dev
ents
invo
lvin
ghu
man
mal
func
tions
.In
the
taxo
nom
yof
PIF,
mai
ndi
stin
ctio
nis
mad
e19
81)
betw
een
PSFs
and
situ
atio
nfa
ctor
s.T
hePS
Fsgr
oup
isag
ain
subd
ivid
edin
tosu
bjec
tive
goal
san
din
tent
ions
,m
enta
llo
adan
dre
sour
ces,
and
affe
ctiv
efa
ctor
s.L
ikew
ise,
situ
atio
nfa
ctor
sar
esu
bdiv
ided
into
task
char
acte
rist
ics,
phys
ical
envi
ronm
ent,
and
wor
ktim
ech
arac
teri
stic
s.T
HE
RP
(Sw
ain
&G
uttm
ann,
1983
)Sw
ain
defin
esPS
F(p
erfo
rman
cesh
apin
gfa
ctor
)si
mpl
yas
fact
ors
that
influ
ence
hum
anpe
rfor
man
ce.
InT
HE
RP,
67PS
Fsar
epr
ovid
ed.
Tho
sear
ecl
assi
fied
into
thre
egr
oups
,i.e
.ex
tern
alPS
Fs,
stre
ssor
sPS
Fsan
din
tern
alPS
Fs.
Aga
in,
the
exte
rnal
PSF
grou
pco
nsis
tsof
thre
esu
bgro
ups:
situ
atio
nal
char
acte
rist
ics
,j
oban
dta
skin
stru
ctio
ns
and
tas
kan
deq
uipm
ent
char
acte
rist
ics
,th
est
ress
ors
PSF
grou
pis
divi
ded
into
psy
chol
ogic
alst
ress
ors
and
phy
siol
ogic
alst
ress
ors
,an
dth
ein
tern
alPS
Fin
clud
eso
rgan
ism
icfa
ctor
s.
PHE
CA
(Wha
lley,
1987
)PH
EC
Aw
asde
velo
ped
for
hum
aner
ror
anal
ysis
inch
emic
alpr
oces
spl
ants
.T
heta
xono
my
isus
edfo
rth
eid
entifi
catio
nof
defic
ient
desi
gnfa
ctor
sca
usin
gsp
ecifi
chu
man
erro
rs.
The
final
set
ofPS
Fsan
dits
stru
ctur
ew
asre
orga
nize
dba
sed
onth
efiv
ePS
Fre
fere
nces
(Sw
ain
and
Gut
tman
n,E
mbr
ey,
Sing
h,A
MA
S,er
gono
mic
liter
atur
e),
vari
ous
inci
dent
/acc
iden
tre
port
sof
chem
ical
pow
erpl
ants
(com
pany
acci
dent
repo
rts,
med
ical
depa
rtm
ent
reco
rds,
inci
dent
repo
rts,
plan
tlo
gbo
oks,
supe
rvis
orre
cord
s,co
mpu
ter
prin
tout
),an
dan
alys
esof
five
even
tca
ses
inw
hich
vari
ous
PIFs
are
inte
rrel
ated
.T
hePS
Fsar
ecl
assi
fied
into
thre
egr
oups
such
asPR
OC
ESS
,PE
RSO
NN
EL
and
ER
GO
NO
MIC
.H
EA
RT
(Will
iam
s,19
88)
HE
AR
Tpr
ovid
es38
EPC
s(e
rror
-pro
duci
ngco
nditi
ons)
asPS
Fs.
EPC
isus
edto
adju
stth
eno
min
alpr
obab
ility
toob
tain
the
final
HE
P.B
ella
my
s(B
ella
my,
1991
)B
ella
my
cate
gori
zes
PSFs
into
eigh
tgr
oups
such
asin
divi
dual
fact
ors,
man
mac
hine
inte
rfac
ech
arac
teri
stic
s(d
ispl
ays
and
cont
rols
),ta
skde
man
ds,
task
char
acte
rist
ics,
inst
ruct
ions
and
proc
edur
es,
stre
sses
,en
viro
nmen
t,so
cio-
tech
nica
lfa
ctor
s.A
sin
TH
ER
P,ST
RE
SSE
Sgr
oup
ofPS
Fsis
sepa
rate
lyor
gani
zed.
Ger
des
(Ger
des,
1997
)PS
Fis
nam
edIF
(infl
uenc
efa
ctor
).In
tota
l,10
8IF
sw
ere
orga
nize
dba
sed
on32
refe
renc
eson
PSF.
All
the
IFs
are
cate
gori
zed
into
four
grou
ps:
HU
MA
N,
TA
SK,
MA
CH
INE
,an
dE
NV
IRO
NM
EN
T.
And
four
grou
psar
eag
ain
divi
ded
into
seve
ral
subg
roup
s.K
-HPE
S(K
EPR
I,19
98)
K-H
PES
was
deve
lope
dfo
ran
alyz
ing
and
repo
rtin
ghu
man
indu
ced/
invo
lved
even
tsin
nucl
ear
pow
erpl
ants
.It
anal
yzes
cogn
itive
erro
rca
uses
and
type
sby
retr
ospe
ctiv
ely
sear
chin
gfo
rin
tern
alde
cisi
on-m
akin
gpr
oces
ses.
Itpr
ovid
es33
inte
rnal
affe
ctin
gfa
ctor
s.P
IFta
xono
mie
sfo
rH
RA
SLIM
(Em
brey
,19
84)
&PL
G-S
LIM
Inth
eor
igin
alSL
IMin
(Em
brey
,19
84),
PSFs
are
sele
cted
thro
ugh
the
expe
rtpa
nel.
The
PSFs
seen
inT
able
2is
the
set
used
asan
exam
ple
in(C
huet
al.,
1994
)(E
mbr
ey,
1984
).T
hese
tof
PIF
sugg
este
din
the
PLG
SLIM
isus
edin
the
eval
uatio
nof
hum
anfa
ilure
even
tsdu
ring
the
low
-pow
eran
dsh
utdo
wn
peri
odof
nucl
ear
pow
erpl
ants
(Chu
etal
.,19
94).
Bot
hth
eor
igin
alSL
IMan
dPL
GSL
IMar
eut
ilize
dto
obta
inth
esu
cces
slik
elih
ood
inde
x(S
LI)
.IN
TE
NT
(Ger
tman
etal
.,19
92)
Ase
tof
PIFs
inIN
TE
NT
isus
edfo
rth
equ
antifi
catio
nof
the
occu
rren
ces
ofer
rors
ofin
tent
ion.
Inad
ditio
n,si
te-s
peci
ficH
EPs
for
erro
rsof
inte
ntio
nar
ede
term
ined
usin
gth
eH
EP
uppe
ran
dlo
wer
boun
dva
lues
.ID
A(P
hilli
ps,
Hum
phre
ys,
PIFs
inID
Aar
ere
pres
ente
din
ahi
erar
chic
alst
ruct
ure
usin
gin
fluen
cedi
agra
m.
HE
Psar
eca
lcul
ated
ina
step
-wis
efa
shio
nto
the
top
hum
anev
ent.
Em
brey
&Se
lby,
1990
)H
RM
S(K
irw
an,
1997
)In
HR
MS,
task
sto
bequ
antifi
edar
eco
mpa
red
with
the
task
with
the
sam
eta
skty
pean
dw
itha
know
nH
EP
invi
ewof
PSF
profi
le.
And
then
,th
eH
EPs
are
mod
ified
acco
rdin
gto
the
diff
eren
ces
inth
epr
ofile
sto
obta
inH
EPs
ofta
sks
tobe
quan
tified
.M
acw
ans
(Mac
wan
&M
osle
h,M
acw
ans
set
ofPI
Fsis
basi
cally
for
the
iden
tifica
tion
ofm
isdi
agno
sis
orer
rors
inin
tent
ion
form
atio
npr
oces
ses.
As
aba
sic
mod
elof
hum
anbe
havi
or19
94)
innu
clea
rpo
wer
plan
ts,
heas
sum
esth
atin
tera
ctio
nsbe
twee
nan
oper
ator
and
apl
ant
occu
ron
the
basi
sof
emer
genc
yop
erat
ing
proc
edur
es(E
OPs
).In
acco
rdan
cew
ithth
em
odel
,th
eco
mpo
sed
set
ofPI
Fsis
com
pris
edof
the
thre
eel
emen
ts,
i.e.
oper
ator
,E
OP
and
plan
t.A
noth
erfe
atur
ein
Mac
wan
sis
that
PIFs
are
clas
sifie
din
toth
esc
enar
io-d
epen
dent
PIFs
that
chan
geas
time
goes
onan
dth
esc
enar
io-i
ndep
ende
ntPI
Fsth
atis
irre
spec
tive
oftim
e.A
ssh
own
inT
able
2,m
ost
PIFs
use
the
term
inol
ogy
that
iscl
osel
yre
late
dw
itha
situ
atio
n.Ju
lius
(Jul
ius,
Jorg
enso
n,Pa
rry
&Ju
lius
reor
gani
zed
the
PIF
set
onth
eba
sis
ofth
eM
acw
ans
taxo
nom
y.In
the
sam
ew
ayas
Mac
wan
s,
PIFs
are
clas
sifie
din
toth
eco
ntex
t-in
depe
nden
tM
osle
h,19
95)
PIFs
and
the
cont
ext-
depe
nden
tPI
Fs.
Eac
hgr
oup
cont
ains
thre
esu
b-gr
oups
ofPI
Fs.
(con
tinu
edon
next
page
)
-
482 J.W. Kim, W. Jung / Journal of Loss Prevention in the Process Industries 16 (2003) 479495T
able
1(c
onti
nued
)
Met
hodo
logy
Des
crip
tion
CR
EA
M(H
olln
agel
,19
98)
Hol
lnag
elpr
ovid
esni
neco
ntex
tfa
ctor
sna
med
CPC
s(c
omm
onpe
rfor
man
ceco
nditi
ons)
.H
esu
gges
tsth
atth
edi
ffer
ence
betw
een
the
CPC
san
dth
eco
nven
tiona
lPS
Fis
that
the
form
erus
edto
adju
stor
prod
uce
the
HE
P,ho
wev
er,
the
latte
ris
used
for
the
over
all
asse
ssm
ent
ofta
sksi
tuat
ion
asw
ell
asth
equ
antifi
catio
nof
HE
P.IN
CO
RE
CT
(Kon
togi
anni
s,19
97)
Kon
togi
anni
sde
fines
PCs
(per
form
ance
cond
ition
s)as
asi
mila
rki
ndof
CPC
s.H
esu
gges
ts,
how
ever
,th
atPC
ssh
ould
beev
alua
ted
atea
chtim
ew
hen
situ
atio
nva
ries
ora
scen
ario
deve
lops
soth
atit
can
beap
plie
dto
the
dyna
mic
risk
asse
ssm
ent
fram
esu
chas
the
dyna
mic
even
ttr
ee(A
cost
a&
Siu,
1993
),in
stea
dof
bein
gas
sess
edon
etim
eat
anea
rly
stag
eof
the
anal
ysis
.H
edi
dno
tin
clud
eco
mpo
site
fact
ors
such
asst
ress
,w
orkl
oad
and
task
com
plex
itysi
nce
thos
ear
eco
mbi
ned
effe
ctof
seve
ral
perf
orm
ance
cond
ition
s.T
aylo
r-A
dam
s(T
aylo
r-A
dam
s,T
heta
xono
my
ofPI
Fth
atT
aylo
r-A
dam
sde
velo
ped
aim
sto
beus
edas
aPS
Fan
alys
ism
odul
ein
CO
RE
-DA
TA
(com
pute
rize
dop
erat
orre
liabi
lity
and
1995
)er
ror
data
base
)w
hich
isa
hum
aner
ror
data
base
for
the
supp
ort
ofH
RA
.C
OR
E-D
AT
Aha
sth
eta
xono
mie
sfo
rfiv
ehu
man
relia
bilit
yre
late
del
emen
ts,
i.e.
exte
rnal
erro
rm
ode,
psyc
holo
gica
ler
ror
mec
hani
sm,
perf
orm
ance
shap
ing
fact
ors,
task
-equ
ipm
ent
taxo
nom
y,an
dhu
man
actio
nta
xono
my.
The
PSF
taxo
nom
yw
asde
velo
ped
base
don
PHE
CA
(Wha
lley,
1987
),T
HE
RP
(Sw
ain
&G
uttm
ann,
1983
),H
EA
RT
(Will
iam
s,19
88).
Rog
ers
(Gib
son
etal
.,19
98)
Aft
erw
ard
Rog
ers
deve
lope
da
new
set
ofPS
Fsfo
rC
OR
E-D
AT
Aco
mpo
sed
of17
PSFs
.A
TH
EA
NA
(US
NR
C,
2000
)A
TH
EA
NA
prov
ides
aco
mpr
ehen
sive
fram
ewor
kco
veri
nger
rors
ofco
mm
issi
on(E
OC
),as
wel
las
erro
rsof
omis
sion
(EO
O).
Itin
trod
uces
the
notio
nof
erro
r-fo
rcin
gco
ntex
t(E
FC)
inw
hich
asi
tuat
ion
iscr
eate
dw
hen
hum
aner
ror
islik
ely
toha
ppen
.A
nE
FCco
mpr
ises
plan
tco
nditi
ons
and
PSFs
.PS
Fsin
clud
etr
aditi
onal
lyus
edte
rms
ofPS
Fssh
own
inT
able
2.
-
483J.W. Kim, W. Jung / Journal of Loss Prevention in the Process Industries 16 (2003) 479495T
able
2T
hePI
Fta
xono
mie
sus
edin
HR
Am
etho
ds
CR
EA
MST
AH
RH
RM
SIN
TE
NT
Ade
quac
yof
orga
niza
tion
Qua
lity
ofin
form
atio
nT
ime
HM
IW
orki
ngco
nditi
ons
Des
ign
Qua
lity
ofin
form
atio
n/in
terf
ace
Stre
ssA
dequ
acy
ofM
MI
and
oper
atio
nal
supp
ort
Mea
ning
fuln
ess
ofpr
oced
ures
Tra
inin
g/E
xper
tise/
Exp
erie
nce/
Com
pete
nce
SRK
Ava
ilabi
lity
ofpr
oced
ures
/pla
nsO
rgan
izat
ion
Proc
edur
esE
xper
ienc
eN
umbe
rof
sim
ulta
neou
sgo
als
Rol
eof
oper
atio
nsT
ask
orga
niza
tion
Safe
tycu
lture
Ava
ilabl
etim
eT
eam
sT
ask
com
plex
ityT
rain
ing
Tim
eof
day
Pers
onal
Mot
ivat
ion
Ade
quac
yof
trai
ning
and
prep
arat
ion
Str
ess
Wor
kloa
dM
oral
e/m
otiv
atio
nSu
perv
isio
nC
ompe
tenc
eC
omm
unic
atio
nPr
oced
ures
SLIM
AT
HE
AN
AIN
CO
RE
CT
Tay
lor-
Ada
ms
Ori
gina
lSL
IMQ
ualit
yof
desi
gnPl
ant
cond
ition
sT
ime
avai
labi
lity
Ala
rms
Mea
ning
fuln
ess
ofpr
oced
ures
Proc
edur
esPl
anav
aila
bilit
yan
dac
cess
ibili
tyC
omm
unic
atio
nR
ole
ofop
erat
ions
Tra
inin
gIn
form
atio
nav
aila
bilit
yan
dac
cess
ibili
tyE
rgon
omic
desi
gnT
eam
sC
omm
unic
atio
nSi
mul
tane
ous
task
sH
MI
ambi
guou
sSt
ress
Supe
rvis
ion
Dec
isio
n-m
akin
gcr
iteri
aH
MI
feed
back
Mor
ale/
Mot
ivat
orSt
affin
gR
espo
nse
dyna
mic
san
dsy
stem
coup
ling
Lab
els
Com
pete
nce
Hum
ans
yste
min
terf
ace
Supe
rvis
ion
Lac
kof
supe
rvis
ion/
chec
ksP
LG
-SL
IMO
rgan
izat
iona
lfa
ctor
sC
apab
ility
degr
adin
gfa
ctor
s(C
DFs
)Pr
oced
ures
Plan
tin
terf
ace
and
indi
catio
nsof
cond
ition
sSt
ress
Tea
mw
ork
and
soci
alfa
ctor
sR
efre
sher
trai
ning
Sign
ifica
ntpr
eced
ing
and
conc
urre
ntac
tions
Env
iron
men
tal
cond
ition
sO
rgan
izat
iona
lfa
ctor
sSt
ress
Tas
kco
mpl
exity
Tas
kco
mpl
exity
Proc
edur
algu
idan
ceT
ask
criti
calit
yT
rain
ing
and
expe
rien
ceT
ask
nove
ltyA
dequ
acy
oftim
eto
acco
mpl
ish
actio
nT
ime
pres
sure
Stre
ssT
rain
ing
Oth
erW
orkl
oad
(con
tinu
edon
next
page
)
-
484 J.W. Kim, W. Jung / Journal of Loss Prevention in the Process Industries 16 (2003) 479495T
able
2(c
onti
nued
)
Rog
ers
Mac
wan
sco
mm
issi
oner
ror
Juliu
sco
mm
issi
oner
ror
Ade
quac
yof
HM
ISc
enar
io-i
ndep
ende
ntPI
FsC
onte
xt-i
ndep
ende
ntPI
FsT
rain
ing
Cre
wtr
aini
ngan
dex
peri
ence
Tra
inin
gre
late
dPr
oced
ures
Cre
wco
nfide
nce
1.D
egre
eof
fam
iliar
ityw
ithan
dfr
eque
ncy
oftr
aini
ngon
,E
OPs
;A
dequ
acy
ofsu
perv
isio
n/m
onito
ring
Rel
ativ
eex
peri
ence
ofR
Oan
dSR
O2.
Gen
eral
philo
soph
yto
war
dsus
ing
the
EO
Ps;
Com
mun
icat
ion
Rec
ent
expe
rien
cew
ithon
eor
mor
efa
ulty
sign
als
3.G
ener
icru
les
for
hand
ling
proc
edur
alam
bigu
ities
;T
eam
orga
niza
tion
Scen
ario
-dep
ende
ntPI
Fs4.
Met
hod
ofre
solv
ing
confl
ictin
gin
form
atio
nfr
omdi
ffer
ent
inst
rum
enta
tions
Stre
ssP
lant
rela
ted
Cre
wte
amch
arac
teri
stic
Tas
kco
mpl
exity
5.V
alue
sof
criti
cal
para
met
ers
5.T
eam
stru
ctur
eT
ask
nove
lty/u
nfam
iliar
ity6.
Rat
eof
chan
geof
criti
cal
para
met
ers
6.E
stab
lishe
dpr
otoc
olfo
rco
mm
unic
atio
nW
orkl
oad
7.In
stru
men
tfa
ilure
7.A
dequ
acy
ofre
sour
ces
Dis
trac
tions
E
OP
rela
ted
Pla
ntre
late
dA
dver
seco
nditi
ons
8.Ph
ase
ofE
OP
8.H
uman
fact
ors
desi
gnof
the
plan
tFa
tigue
9.T
ype
oflo
gic
stru
ctur
eC
onte
xt-d
epen
dent
PIFs
Mot
ivat
ion
10.
Num
ber
oflo
gica
lco
nditi
ons
Pla
ntre
late
dSa
fety
cultu
re
Ope
rato
rre
late
d9.
Val
ueof
criti
cal
para
met
erA
dequ
acy
ofde
sign
11O
pera
tor
diag
nosi
s10
.T
rend
ofcr
itica
lpa
ram
eter
sR
obus
tnes
sof
desi
gn12
Mem
ory
ofre
cent
actio
ns11
.A
vaila
bilit
yof
equi
pmen
t13
Perc
eive
dim
port
ance
12.
Ava
ilabi
lity
ofin
stru
men
tatio
n14
Perc
eive
dco
nseq
uenc
esE
OP
rela
ted
15O
pera
tor
expe
ctat
ions
13.
EO
Pre
spon
seph
ase
(ver
ifica
tion,
diag
nosi
s)O
pera
tor
rela
ted
14.
Con
fiden
cein
diag
nosi
s15
.E
xpec
tatio
n16
.M
emor
yof
prev
ious
actio
nsan
dac
cide
nthi
stor
y
-
485J.W. Kim, W. Jung / Journal of Loss Prevention in the Process Industries 16 (2003) 479495
2.2. Development of a new full-set of PIF
Besides the 18 taxonomies described in the previoussection, other literature dealing with important influenc-ing factors related to team/organization behaviors havealso been included in the review. For instance, Jacobsand Haber (1994) describe organizational factors thatcan affect system safety, and Hollnagel (1998); Salas,Dickinson, Converse and Tannenbaum (1992); Xiao,Hunter, Mackenzie, Jefferies and Horst (1996), andUrban, Weaver, Bowers and Rhodenizer (1996) dealwith inter/intra team interaction. In addition to those,HRA practitioners opinions were also utilized to deriveimportant factors. Especially, factors related to systemdynamic features were much supplemented.
All the factors collected from the above-mentionedsources were collated into a new full-set PIF taxonomy.It is assumed that the context under which an operatingcrew should perform given tasks can be modeled as inFig. 1. According to the Fig. 1, the collated PIFs areclassified into four main groups: HUMAN, SYSTEM,TASK, and ENVIRONMENT. The boundary of eachgroup is defined as follows.
HUMAN: Personal characteristics and working capa-bilities of the human operator.
SYSTEM: MMI, plant hardware system, and physicalcharacteristics of the plant process.
TASK: Procedures and task characteristics requiredof the operator.
ENVIRONMENT: Team and organization factors,and physical working environment.
The four main groups are again divided into severalsubgroups. The final full set PIF taxonomy collated intothe new classification frame is shown in Table 3.
2.3. Principal trend in the use of PIF for HRA
This section presents the trend of the selection and thelevel of definition of PIFs for use in the HRA, as HRAmethodology develops.
Fig. 1. The task context of nuclear power plants.
Firstly, as HRA methodology has improved, variouskinds of factors have been considered. In early HRAmethods such as THERP, ASEP and HCR, a very limitednumber of PIFs were chosen. For the methods utilizingexpert judgment, such as SLIM and STAHR, team andorganization factors , in addition to individual factors ,could be included. In INTENT, safety culture was con-sidered as one of the important PIFs. Among the recentlydeveloped HRA methods, CREAM and INCORECTconsider team and organization factors , systemcharacteristics , and simultaneous goals/tasks asimportant factors for human error/reliability analysis,and furthermore, the meaning of PIFs becomes moreclearly defined than in the previous quantitative HRAmethods. ATHEANA suggests the plant conditions,including system dynamic conditions, as the constituentsthat create a situation in which human error is stronglylikely to happen, i.e. error-forcing context (EFC). Tosummarize the trend in the selection of PIFs for use inthe HRA, in the initial HRA methods, a few very limitedfactors such as individual factors , procedures , andMMI factors were considered in evaluating humanreliability. Gradually, as HRA method becomes sophisti-cated, the assessment of overall work context is emphas-ized for better analysis and prediction of human error.According to this trend, team and organization factors ,safety culture , plant dynamic features , and simul-taneous tasks/goals appeared as important factors whichshould be assessed.
Secondly, PIFs could be grouped as follows withrespect to the frequency they are used in HRA methods.
Factors that are used in the majority of HRA methods:training, experience, procedure, MMI/information,and time.
Factors that are moderately used in HRA methods:stress, workload, motivation, task complexity, simul-taneous goals/tasks, working condition, supervision,team factors, and communication.
Factors that are used in a minority of HRA methods:adequacy of resources, decision making criteria,response dynamics and system coupling, availabilityof equipment, trend and value of critical parameters,time of day, organization factors, task organization,and safety culture.
Thirdly, the terminology and meanings of the selectedPIFs became more specific and practical. Some of thefactors being used in HRA methods are comprehensivein the meaning of their terminology. For this reason, theassessment results between analysts could be differentbecause they assess the factors with their subjectivemeaning. Such factors, for instance, include stress, work-load, task complexity, safety culture, organization factor,etc. For such factors, it is recommended that more appar-ent terminology be used or that the meaning/assessment
-
486 J.W. Kim, W. Jung / Journal of Loss Prevention in the Process Industries 16 (2003) 479495T
able
3T
heFu
ll-se
tPI
FT
axon
omy
Mai
ngr
oup
Subg
roup
Det
aile
dite
ms
HU
MA
NC
ogni
tive
char
acte
rist
ics
Cog
nitiv
est
ates
Tem
pora
lco
gniti
vest
ates
atte
ntio
nm
emor
yof
rece
ntac
tions
int
ellig
ence
ope
rato
rdi
agno
sis
ski
llle
vel
per
ceiv
edim
port
ance
kno
wle
dge
per
ceiv
edco
nseq
uenc
ese
xper
ienc
eo
pera
tor
expe
ctat
ions
tra
inin
gc
onfid
ence
indi
agno
sis
mem
ory
ofpr
evio
usac
tions
and
acci
dent
hist
ory
Phys
ical
and
psyc
holo
gica
lch
arac
teri
stic
sPh
ysic
alst
ates
Psyc
holo
gica
lst
ates
gen
der/
age
em
otio
n/fe
elin
gm
otor
skill
sc
onfu
sion
/per
plex
ityp
hysi
cal
disa
bilit
ies
tas
kbu
rden
im
pedi
men
t:si
ght/h
eari
ng/s
peak
ing
fea
rof
failu
re/c
onse
quen
ces
cla
rity
insp
eaki
ng/u
seof
stan
dard
lang
uage
hig
hje
opar
dyri
skf
atig
ue/p
ain
dis
com
fort
hun
ger,
thir
st
Pers
onal
and
soci
alch
arac
teri
stic
sPe
rson
alSo
cial
atti
tude
sta
tus
mor
ale/
mot
ivat
ion
rol
e/re
spon
sibi
lity
ris
kta
king
nor
ms
sel
f-es
teem
and
self
-con
fiden
cea
ttitu
des
base
don
influ
ence
sof
fam
ilyan
dot
her
sen
seof
resp
onsi
bilit
you
tsid
epe
rson
sor
agen
cies
sen
satio
nse
ekin
gl
eade
rshi
pab
ility
soc
iabi
lity
per
sona
lity
ant
icip
atio
n(c
onti
nued
onne
xtpa
ge)
-
487J.W. Kim, W. Jung / Journal of Loss Prevention in the Process Industries 16 (2003) 479495T
able
3(c
onti
nued
)
Mai
ngr
oup
Subg
roup
Det
aile
dite
ms
TA
SKPr
oced
ures
ava
ilabi
lity
lev
elof
stan
dard
izat
ion
inus
eof
term
inol
ogy
for
mat
orty
ped
ecis
ion
mak
ing
crite
rion
qua
lity
log
icst
ruct
ure
lev
elof
deta
iln
umbe
rof
logi
cal
cond
ition
s(b
ranc
hes)
int
erpr
etat
ion
mar
gin
num
ber
ofsi
mul
tane
ous
task
sn
umbe
rof
step
sa
dequ
acy
ofca
utio
n/w
arni
ngr
equi
red
time
for
com
plet
ion
cla
rity
ofin
stru
ctio
nan
dte
rmin
olog
y
Tas
kch
arac
teri
stic
sT
ask
type
typ
eof
man
mac
hine
inte
ract
ion:
EO
Pph
ase
pr
oced
ure
follo
win
g,m
onito
ring
,de
tect
ion,
veri
ficat
ion,
diag
nosi
s,re
cove
ryr
equi
red
leve
lof
cogn
ition
dyn
amic
vs.
step
-by-
step
activ
ities
Tas
kat
trib
ute/
requ
irem
ent
num
ber
ofre
quir
edin
form
atio
nn
umbe
rof
nece
ssar
yin
form
atio
nto
bem
emor
ized
inf
orm
atio
nlo
adt
ask
diffi
culty
m
ultip
leat
tem
pts
nar
row
ness
tas
kno
velty
tas
kco
nseq
uenc
esd
egre
eof
loca
lop
erat
ion
fre
quen
cyan
dfa
mili
arity
ofta
skd
egre
eof
disc
repa
ncy
with
fam
iliar
task
sn
umbe
rof
sim
ulta
neou
sgo
als/
task
sc
oncu
rren
tac
tiviti
esan
din
terr
uptio
nsi
nter
rupt
ion
from
othe
rpe
rson
nel
dis
crep
ancy
betw
een
trai
ning
and
real
ityn
eces
sity
ofau
xilia
ryto
ols
con
flict
sof
mot
ives
abou
tjo
bpe
rfor
man
cea
ppro
pria
tene
ssof
requ
ired
tool
sm
ultip
lese
nsor
yre
quir
emen
tsp
erce
ptua
lre
quir
emen
tst
ask
criti
calit
yp
hysi
cal
requ
irem
ents
deg
ree
ofm
anua
lop
erat
ion
mot
orre
quir
emen
tsm
uscu
lar
pow
ers
peed
dex
teri
typ
reci
sion
-
488 J.W. Kim, W. Jung / Journal of Loss Prevention in the Process Industries 16 (2003) 479495T
able
3(c
onti
nued
)
Mai
ngr
oup
Subg
roup
Det
aile
dite
ms
cal
cula
tiona
lre
quir
emen
ta
ntic
ipat
ory
requ
irem
ent
req
uire
men
ton
and
type
offe
edba
ckd
egre
eof
refe
renc
eto
othe
rm
ater
ials
besi
depr
oced
ures
com
mun
icat
ion
requ
irem
ent
tea
mco
oper
atio
nre
quir
emen
tn
eces
sity
ofde
cisi
onm
akin
gfr
omhi
gher
orga
niza
tion
SYST
EM
MM
IIn
dica
tor/
cont
rolle
rsPa
nel/s
cree
nla
yout
ava
ilabi
lity
rea
chab
ility
rel
iabi
lity
vis
ibili
tyd
iscr
imin
atio
n/di
stin
guis
habi
lity
ofsi
gnal
sc
odin
g/la
belin
ga
ttrib
utes
com
patib
ility
typ
eof
sens
ing
sta
teof
arra
ngem
ent
con
trol
dis
play
rela
tions
hips
com
plic
ated
ness
ofM
CR
pane
lo
rien
tatio
nSu
ppor
tsy
stem
sl
abel
ing
ava
ilabi
lity/
adeq
uacy
loc
atio
na
vaila
bilit
y/ad
equa
cyof
spec
ial
equi
pmen
t,to
ols
exi
sten
ceof
faile
din
stru
men
tan
dsu
pplie
se
xist
ence
offa
iled
indi
cato
ru
sabi
lity
ofre
quir
edfu
nctio
ns
tuck
inst
rum
ent
con
flict
ing
sign
als/
cues
Syst
emst
ates
inh
eren
tsy
stem
com
plex
ityo
rgan
izat
ion
ofco
mpo
nent
sn
umbe
rof
coup
led
com
pone
nts
rel
iabi
lity
red
unda
ncy
lev
elof
auto
mat
ion
con
figur
atio
nof
syst
em(o
n/of
fst
ate
ofco
mpo
nent
s)s
tuck
/fai
led
com
pone
nts
pre
viou
sop
erat
ion
ofco
mpo
nent
and
curr
ent
oper
atin
gst
atus
ava
ilabi
lity
ofvi
tal
sour
ces
Phen
omen
olog
ical
char
acte
rist
ics
Phys
ical
char
acte
rist
ics
Ope
ratio
nal
char
acte
rist
ics
rat
eof
chan
geof
criti
cal
para
met
ers
sud
denn
ess
ofon
set
tre
ndof
criti
cal
para
met
ers
ove
rlap
with
prev
ious
task
sv
alue
ofcr
itica
lpa
ram
eter
st
otal
time
toC
Dfr
omth
ein
itiat
ion
ofac
cide
ntr
equi
red
wat
erle
vel
vs.
curr
ent
stat
usof
wat
erle
vel
tim
eav
aila
ble
for
oper
ator
perf
orm
ance
num
ber
ofdy
nam
icch
angi
ngva
riab
les
(pla
ntdy
nam
icbe
havi
or)
tim
epr
essu
re(t
ime
requ
ired
vs.
time
avai
labl
e)h
ighl
yun
stab
lepl
ant
situ
atio
n
unce
rtai
nin
form
atio
ne
xist
ence
ofpr
eced
ing
info
rmat
ion
onsc
enar
iop
hase
ofpl
ant
oper
atio
nE
OP
resp
onse
phas
e(v
erifi
catio
n,di
agno
sis)
deg
ree
ofal
arm
aval
anch
e
-
489J.W. Kim, W. Jung / Journal of Loss Prevention in the Process Industries 16 (2003) 479495T
able
3(c
onti
nued
)
Mai
ngr
oup
Subg
roup
Det
aile
dite
ms
Env
iron
men
tPh
ysic
alw
orki
ngco
nditi
ons
Phys
ical
cons
trai
nts
Tim
ing
aspe
cts
con
tam
inat
ion
from
radi
oact
ive
mat
eria
lt
ime
ofda
yp
hysi
cal
inco
nven
ienc
efr
ompr
otec
tive
clot
hing
tim
eon
duty
tem
pera
ture
/hum
idity
/pre
ssur
e/ill
umin
atio
nt
ime
into
scen
ario
int
erfe
renc
ein
com
mun
icat
ion
cir
cadi
anrh
ythm
effe
cts
noi
sev
ibra
tion
G-f
orce
sa
irpr
essu
re/q
ualit
y/ve
ntila
tion/
oxyg
enin
suffi
cien
cym
ovem
ent
cons
tric
tion
mov
ing
dist
ance
nar
row
wor
ksp
ace
orob
stac
les
dan
gero
usw
ork
spac
ee
leva
ted
wor
ksp
ace
acc
essi
bilit
yof
com
pone
nts
arc
hite
ctur
alfe
atur
eso
rder
and
clea
nlin
ess
Tea
man
dor
gani
zatio
nfa
ctor
sT
eam
-rel
ated
fact
ors
Tra
inin
gc
lear
ness
injo
bde
scri
ptio
nor
role
defin
ition
sim
ulat
ion
scop
eof
sim
ulat
orc
lear
ness
inre
spon
sibi
litie
s/co
mm
unic
atio
nlin
efi
delit
yof
sim
ulat
ion
scen
ario
ade
quac
yof
dist
ribu
ted
wor
kloa
df
requ
ency
and
trai
ning
time
int
ra/in
ter-
team
coop
erat
ion
tim
epe
riod
betw
een
trai
ning
sess
ions
tea
mco
hesi
vene
ss/c
olla
bora
tion
Man
agem
ent
and
polic
ya
bilit
y/le
ader
ship
/aut
hori
tyof
team
lead
erw
ork
(tas
k)or
gani
zatio
nc
omm
itmen
tto
lead
ersh
ipw
ork/
rest
sche
dule
Tea
mco
mm
unic
atio
n-re
late
dfa
ctor
ss
hift
orga
niza
tion
str
uctu
reof
inst
ruct
ion/
info
rmat
ion
deliv
ery
shi
ftro
tatio
ns
tand
ardi
zatio
nin
inst
ruct
ion/
info
rmat
ion
deliv
ery
sup
port
ing
team
sta
ndar
dco
mm
unic
atio
nst
ruct
ure/
netw
ork
lev
elof
supe
rvis
ion
med
iaof
inst
ruct
ion/
info
rmat
ion
deliv
ery
(pag
eph
one,
fax,
pape
r,et
c.)
ina
dequ
ate
inst
ruct
ion
est
ablis
hed
prot
ocol
/for
mof
inst
ruct
ion/
info
rmat
ion
deliv
ery
(int
ra-
pla
ntpo
licy
team
/inte
r-te
am,
oral
/wri
tten,
form
al)
mai
nten
ance
pro
cedu
res
ofin
stru
ctio
n/in
form
atio
nde
liver
yh
uman
reso
urce
deve
lopi
ngp
roto
col
betw
een
send
eran
dre
ceiv
eri
nves
tmen
tsan
dcu
tsq
ualit
yas
sura
nce
saf
ety
mea
sure
sr
ewar
dsan
dpu
nish
men
tsw
ork
met
hods
man
ning
Safe
tycu
lture
rou
tine
viol
atio
nss
afet
y/ec
onom
ytr
adeo
ffo
penn
ess
inco
mm
unic
atio
n
-
490 J.W. Kim, W. Jung / Journal of Loss Prevention in the Process Industries 16 (2003) 479495
items be clearly defined so that the consistency betweenanalysts is maintained.
3. Situational characteristics during accidentmanagement
This section describes the situational characteristicsduring accident management (including emergencyoperation) in nuclear power plants. Based on thosedescriptions, performance influencing elements thataffect operator performance during accident manage-ment are determined. Behavioral, phenomenological,organizational and environmental characteristics havebeen examined based on the literature dealing with acci-dent management in nuclear power plants (KEPRI, 1997;Ha et al., 1997; US NRC, 1988).
Firstly, operators tasks required in accident manage-ment (AM) situations are composed of not simplemanipulative actions, but cyclic cognitive activities suchas detection, observation, diagnosis, evaluation, plan-ning, and decision-making. In performing such activities,
Table 4Sample tasks appearing in accident management guidance (adapted from WOG, 1994)
1. Evaluate the following negative impacts associated with implementing the strategy.Check the following two conditions, and if both conditions are satisfied, there will be insufficient injection source .Check if core has not been reflooded.Check the following condition: RWST level (L09)%.2. Evaluate mitigating actions.If there is a possibility of insufficient containment injection source, then evaluate the following mitigating actions.Evaluate the action of Increase RWST refill rateEvaluate the action of Control containment injection flowrate to maintain RWST water level greater than (L09)% .3. Evaluate the consequences of NOT injecting into the containment (Containment challenge due to Basemat Meltthrough, RPV failure maynot be delayed, Fission products from ex-vessel core debris, Consequences of HPME, Combustible gas generation due to CCI, Recirculationproblem)4. Determine if containment injection should be initiated by comparing the consequences of NOT injecting into the containment versus thenegative impacts of injecting into the containment.
Table 5The situational characteristics of AM tasks and related human factors
Situational characteristics under AM Human factors
Behavioral characteristics Cognitive activities (observation, diagnosis, Knowledge (training, experience)interpretation, planning, decision-making, etc.) Procedure
Manmachine interface
Phenomenological characteristics Dynamic evolution of plant systems Dynamic system statusMultiple events Multiple goals and tasks
Available time for task achievement
Organizational characteristics Inter-/Intrateam cooperation Team and organization factorsDecision-making from higher organizations Decision-making structure
Plant policy and safety culture
Environmental characteristics Local operation Task location and physical environmentHarsh environment
procedures, information, operator support systems, train-ing and experiences are crucial for the proper assessmentof plant dynamic situations and the appropriate planningof how to cope with the given situation. Some exampletasks appearing in accident management guidance areprovided in Table 4.
Secondly, as an accident scenario evolves, plant sys-tems and physical phenomena also change dynamicallyand multiple events can take place. Those complex and
Fig. 2. Emergency organizations and interactions between organiza-tions in nuclear power plant emergency situations.
-
491J.W. Kim, W. Jung / Journal of Loss Prevention in the Process Industries 16 (2003) 479495
Tab
le6
Can
dida
tePI
Fsse
lect
edco
nsid
erin
gA
Msi
tuat
iona
lch
arac
teri
stic
s
HU
MA
NT
ASK
SYST
EM
EN
VIR
ON
ME
NT
kno
wle
dge,
expe
rien
ce,
trai
ning
,co
mpe
tenc
ePr
oced
ures
MM
I/In
form
atio
nW
orki
ngco
nditi
ons
tres
s,bu
rden
,fe
arof
cons
eque
nces
Pro
cedu
res
sys
tem
stat
esp
hysi
cal
cons
trai
nts
(for
loca
lac
tiviti
es)
atti
tude
,m
oral
e,m
otiv
atio
nt
rend
ofcr
itica
lpa
ram
eter
st
ime
ofda
y(c
irca
dian
rhyt
hm)
ris
k-ta
king
,co
nfide
nce
val
ueof
criti
cal
para
met
ers
num
ber
ofdy
nam
icch
angi
ngva
riab
les
deg
ree
ofal
arm
aval
anch
es
udde
nnes
sof
onse
tt
ime
adeq
uacy
Tas
kty
peT
eam
fact
ors
typ
eof
man
-mac
hine
inte
ract
ion
tea
mst
ruct
ure
lev
elof
cogn
ition
rol
e/re
spon
sibi
lity
-dyn
amic
/ste
p-by
-ste
pac
tiviti
esa
dequ
acy
ofdi
stri
bute
dw
orkl
oad
tea
mco
oper
atio
nan
dco
llabo
ratio
nc
omm
unic
atio
nT
ask
attr
ibut
e/re
quir
emen
tO
rgan
izat
ion
fact
ors
tas
kcr
itica
lity
wor
kor
gani
zatio
nt
ask
nove
ltys
uper
visi
ont
ask
cons
eque
nces
shi
ftor
gani
zatio
nd
egre
eof
loca
lop
erat
ion
pla
ntpo
licy
cal
cula
tiona
lre
quir
emen
ta
dequ
acy
ofin
stru
ctio
na
ntic
ipat
ory
requ
irem
ent
mai
nten
ance
com
mun
icat
ion
requ
irem
ent
saf
ety
cultu
ren
eces
sity
ofde
cisi
on-m
akin
gfr
omhi
gher
org
aniz
atio
nst
ruct
ure
and
man
ning
orga
niza
tion
nec
essi
tyof
auxi
liary
tool
sn
umbe
rof
nece
ssar
yin
form
atio
nd
egre
eof
refe
renc
eto
othe
rpr
oced
ures
deg
ree
ofdi
scre
panc
yw
ithfa
mili
arta
sks
sim
ulta
neou
sgo
als/
task
s
-
492 J.W. Kim, W. Jung / Journal of Loss Prevention in the Process Industries 16 (2003) 479495
Fig. 3. Process of structuring PIFs relevant to AM situations.
Table 7An example of structuring sub-itemsprocedures
Full-set PIFs Sub-items
TASK TASKProcedures Proceduresavailability format or type qualitylevel of detailmargin on interpretationclarity of instruction and terminology level of standardization in use of terminologydecision-making criterion logic structure number of logical conditions (branches)number of simultaneous tasksadequacy of caution/warning
dynamic situations can be a critical influencing factorto operator performance. Multiple tasks, status of majorsystems and components, and temporal factors arerelated to this situational characteristic.
Thirdly, most of AM tasks require inter-/intra-teamcooperation and decision-making in higher-level organi-zations beyond individual execution as in normal oper-ation. Under those situations, various factors related toteam performance, decision-making structure in a teamor organization, and plant policy and safety culture can
be important contributors to task performance. Fig. 2shows the emergency organizations and interactionsbetween those organizations in nuclear power plants.
Fourthly, harsh working environments such as highradioactivity, high pressure/temperature, and potential ofhydrogen detonation, etc., may take place in some tasklocations. Such environmental features can be a directobstacle, i.e. limitation in accessibility, to task perform-ance.
Table 5 summarizes the situational characteristics ofAM tasks and related human factors. Candidate PIFs forstructuring to be fit for HRA are extracted from the full-set PIF taxonomy considering those situational charac-teristics and human factors relevant to AM. The list ofextracted candidate PIFs is shown in Table 6.
4. Structuring PIFs for human reliability analysis
The candidate PIFs for AM obtained in the previoussection should be reorganized to be appropriate for usein HRA. The structuring of PIFs was conducted on thebasis of the insights from the Section 2.3 Principal trendin the use of PIF for HRA. Fig. 3 shows the schematicrepresentation of the process to the construction of a setof PIFs for AM HRA. The PIFs relevant to AM situ-ations have firstly been selected from the full-set PIFtaxonomy, and those PIFs are grouped into ones with
-
493J.W. Kim, W. Jung / Journal of Loss Prevention in the Process Industries 16 (2003) 479495
Table 8A set of PIFs for human error analysis of AM tasks
PIF group Representative PIF Subitems
HUMAN 1. Training and experience Adequacy of training (frequency, recent training, fidelity of simulation program)Experiences/practices of real operating eventsLearning of the past events/experiencesCareer of the operators
TASK 2. Availability and quality of procedures AvailabilityFormat or typeClarity of instruction and terminologyDecision-making criterionLogic structure
3. Simultaneous goals/tasks Number of simultaneous goals/tasksPriority between goals/tasksConflict between goals
4. Task type/attributes Type of manmachine interactionDynamic/step-by-stepTask criticality/consequencesDegree of discrepancy with familiar tasks
SYSTEM 5. Availability and quality of information Information availability (instrumentation fail/stuck)Clearness of meaning (direct indication/interpretationrequired/ambiguous/unreliable information)Distinguishability of informationControl display relationships
6. Status and trend of critical parameters Value of critical parametersTrend of critical parameters (rate of change of critical parametersNumber of dynamic changing variablesDegree of alarm avalanche
7. Status of safety system/component Success/fail state of safety system/componentLevel of trust on the system/componentNumber of failed/stuck componentsPrevious operation history and current status of safety system
8. Time pressure Available time vs. required time
ENVIRONMENT 9. Working environmental features Task location: (MCR/local CR/local area)Accessibility
10. Team cooperation and communication Clearness in role/responsibility definitionDirection, type, method, protocolStandardization in instruction/information deliveryTeam collaboration/cooperationAdequacy of distributed workload
11. Plant policy and safety culture Plant specific prioritized (or preference for/objection to) goals/strategiesAttitude toward EOP/AMP trainingSafety/economy tradeoffRoutine violations
similar meaning. Then, considering the situationalcharacteristics of AM tasks and the criteria for use inHRA, the final taxonomy is systematized. The criteriafor the selection and structuring of PIFs for HRA areestablished as follows, considering the practical aspectsof HRA.
It should include all the important factors to assessoverall task context as comprehensively as it can.
It should not be overlapped between factors. It gave the keynote for selecting factors that directly
affect human error occurrences.
-
494 J.W. Kim, W. Jung / Journal of Loss Prevention in the Process Industries 16 (2003) 479495
It selects the factors that could be reflected intohuman error/reliability analysis.
It selects the factors that are assessable in practice. The terminology should be as specific and practical
as possible so that the meanings of the terms mightbe easily understood.
According to these criteria, construction of the rep-resentative PIFs and their subitems was made using oneor mixed method of the following structuring processesby the property of each PIF. The first structuring methodis to use the full-set PIF taxonomy illustrated in Table7. Important subitems for the assessment of a PIF aremarked out from the given detailed items of a PIF. Thesecond method is to choose the representative PIF fromthe PIF group with similar meaning and to organize subi-tems. Thirdly, other factors from the literature and HRAexperts were also incorporated into those processes.
Through repetitive modification, the final set of PIFs,which is composed of 11 representative PIFs and theirsubitems, is presented in Table 8. The proposed tax-onomy will need a continual revision process in associ-ation with the methodology development.
5. Conclusions
In this paper, we introduced a process of the system-atic construction of PIF sets for the use in HRA in aspecific context, especially, AM situations (includingemergency operations) in nuclear power plants. In orderto achieve this goal, firstly, 18 existing PIF taxonomiesincluding the recently developed methods have been col-lected and collated into a new set of PIFs consisting ofapproximately 220 detailed PIFs. In particular, PIF taxo-nomies for HRA were analytically reviewed in view ofthe trend and characteristics of selection and use, andthe insights from the review were considered in thedevelopment of the new taxonomy. In order to reflectthe task context, situational characteristics during theAM realm in nuclear power plants were describedaccording to the behavioral, phenomenological, organi-zational and environmental characteristics. Based on thedescriptions, PIFs relevant to such situational character-istics were selected from the full-set PIFs and, finally,those were structured in a two-layer hierarchical formconsidering the practical aspects of the use of PIFs inHRA.
The proposed taxonomy could be used in the assess-ment of overall task context, prediction of human error,such as internal/external error modes, and quantificationof their probabilities, and so on. In order to use the tax-onomy in those areas, an appropriate analysis/assessmentframework should be developed. For example, thepsychological error mechanisms (PEMs) in human cog-nitive functions may help find a way to consider the PIFs
in a systematic manner. In accordance with the PEM,some PIFs may be directly causing factors and othersbe contributory or shaping factors to human error. Theproposed taxonomy would require continual modifi-cations in association with the development ofHEA/HRA methodology.
Acknowledgements
This research has been carried out under the NuclearR&D Program by MOST (Ministry of Science andTechnology), Republic of Korea.
References
Acosta, C., & Siu, N. (1993). Dynamic event trees in accident sequenceanalysis: application to steam generator tube rupture. ReliabilityEngineering and System Safety, 41, 135154.
Bellamy, L. J. (1991). The quantification of human fallibility. Journalof Health and Safety, 6, 1322.
Chu, T. L., Musicki, Z., Yang, J., Kohut, P., Bozoki, G., Hsu, C. J.,Diamond, D. J., Wong, S. M., Upton, N. Y., Bley, D., & Johnson,D. (1994). Evaluation of potential severe accidents during lowpower and shutdown operations at Surry Unit 1. NUREG/CR-6144,2, (1B), USNRC.
Dougherty, E. D. (1993). Context and human reliability analysis.Reliability Engineering and System Safety, 41, 2547.
Embrey, D. (1984). SLIM-MAUD: An approach to assessing humanerror probabilities using structured expert judgement. NUREG/CR-3518, USNRC.
Gerdes, G. (1997). Identification and analysis of cognitive errors:application to control room operators. PhD thesis.
Gertman, D. I., Blackman, H. S., Haney, L. N., Seidler, K. S., & Hahn,H. A. (1992). INTENT: A method for estimating human error prob-abilities for decision-based errors. Reliability Engineering and Sys-tem Safety, 35, 127136.
Gibson, H., Basra, G. & Kirwan, B. (1998). Development of theCORE-DATA database. The final IAEA-RCM on collection andclassification of human reliability data for use in probabilisticsafety assessments, IAEA-J4-RC589.3B.
Ha, J., Kim, J. W., Jae, M., Yu, D., Park, R. J., Jeong, K. S., Suh, K.Y., & Park, C. K. (1997). Development of accident managementtechnology and computer codes, KAERI/RR-1742/96. Taejon,Korea: KAERI.
Hannaman, G. W., Spurgin, A. J., & Lukic, Y. D. (1984). Humancognitive reliability model for PRA analysis (draft report). EPRIRP-2170-3.
Hollnagel, E. (1998). Cognitive reliability and error analysis method-ology. London: Elsevier.
Hollnagel, E., Kaarstad, M., & Lee, H. C. (1999). Error mode predic-tion. Ergonomics, 42(11), 14571471.
Jacobs, R., & Haber, S. (1994). Organizational processes and nuclearpower plant safety. Reliability Engineering and System Safety, 45,7583.
Julius, J., Jorgenson, E., Parry, G. W., & Mosleh, A. M. (1995). Aprocedure for the analysis of errors of commission in a probabilisticsafety assessment of a nuclear power plant at full power. ReliabilityEngineering and System Safety, 50, 189201.
KEPRI (1997). Preliminary study for development of accident manage-ment plans in nuclear power plants. TR.96NJ11.97.77, Taejon,Korea.
-
495J.W. Kim, W. Jung / Journal of Loss Prevention in the Process Industries 16 (2003) 479495
KEPRI (1998). Development of Korean HPES (human performanceenhancement system) for nuclear power plants.TR.95ZJ04.J1998.21, Taejon, Korea.
Kim, J. W., & Ha, J. (1997). The evaluation of accident managementstrategy involving operator action. Journal of the Korean NuclearSociety, 29(5), 368374.
Kirwan, B. (1992a). Human error identification in human reliabilityassessment, part 1: overview of approaches. Applied Ergonomics,23, 299318.
Kirwan, B. (1992b). Human error identification in human reliabilityassessment, part 2: detailed comparison of techniques. AppliedErgonomics, 23, 371381.
Kirwan, B. (1997). The development of a nuclear chemical planthuman reliability management approach: HRMS and JHEDI.Reliability Engineering and System Safety, 56, 107133.
Kontogiannis, T. (1997). A framework for the analysis of cognitivereliability in complex systems: a recovery centred approach.Reliability Engineering and System Safety, 58, 233248.
Macwan, J., & Mosleh, A. (1994). A methodology for modeling oper-ator errors of commission in probabilistic risk assessment.Reliability Engineering and System Safety, 45, 139157.
Moieni, P., Spurgin, A. J., & Singh, A. (1994). Advances in humanreliability analysis methodology. part I: frameworks, models anddata. Reliability Engineering and System Safety, 44, 2755.
Nakanishi, M., Konishi, T., & Okada, Y. (2002). A study on analysismethod of performance shaping factors on human error. In The3rd International Forum on Safety Engineering and Science (IFSESIII), Nihon University, Japan.
Paradies, M., Unger L., Haas, P., Terranova, M. (1993). Developmentof the NRCs human performance investigation process (HPIP).NUREG/CR-5455, Vol. 1, USNRC.
Phillips, L. D., Humphreys, P., Embrey, D., & Selby, D. L. (1990). Asocio-technical approach to assessing human reliability. In R. M.Oliver, & J. Q. Smith (Eds.), Influence diagrams, belief nets anddecision analysis. John Wiley & Sons.
Rasmussen, J., Pedersen, O. M., Mancini, G., Carnino, A., Griffon,M., & Gagnolet, P. (1981). Classification system for reporting
events involving human malfunctions, RISO-M-2240. Denmark:RISO National Laboratory.
Salas, E., Dickinson, T. L., Converse, S. A., & Tannenbaum, S. I.(1992). Toward an understanding of team performance and train-ing. In R. W. Swezey, & E. Salas (Eds.), Teams, their training andperformance (pp. 329). Norwood: Albex Publishing.
Swain, A., & Guttmann, H. E. (1983). Handbook of human reliabilityanalysis with emphasis on nuclear power plant applications,NUREG/CR-1278. USA: US NRC.
Taylor-Adams, S. E. (1995). An overview of the development of thecomputerized operator reliability and error database (CORE-DATA). The first IAEA-RCM on collection and classification ofhuman reliability data for use in probabilistic safety assessments,IAEA-J4-RC589.
Urban, J. M., Weaver, J. L., Bowers, C. A., & Rhodenizer, L. (1996).Effects of workload and structure on team processes and perform-ance: implications for complex team decision-making. Human Fac-tors, 38, 300310.
US NRC (1988). Integration plan for closure of severe accident issues.SECY-88-147, USA.
US NRC (1989). Staff plans for accident management regulatory andresearch program, SECY-89-012, USA.
US NRC (2000). Technical basis and implementation guidelines for atechnique for human event analysis (ATHEANA). NUREG-1624Rev. 1.
Xiao, Y., Hunter, W. A., Mackenzie, C. F., Jefferies, N. J., & Horst,R. L. (1996). Task complexity in emergency medical care and itsimplications for team coordination. Human Factors, 38(4), 636645.
Whalley, S. P. (1987). Factors affecting human reliability in the chemi-cal process industry. PhD thesis, Aston University.
Williams, J. C. (1988). A data-based method for assessing and reduc-ing human error to improve operational performance. In Proceed-ings of the IEEE Fourth Conference on Human Factors and PowerPlants, Monterey, California. (pp. 436450).
WOG (1994). Severe accident management guidance, vol. 2: guide-lines. WOG-SAMG, Westinghouse Electric Corporation.
A taxonomy of performance influencing factors for human reliability analysis of emergency tasksIntroductionCollection and analysis of the existing PIF taxonomiesReview of the existing PIF taxonomiesDevelopment of a new full-set of PIFPrincipal trend in the use of PIF for HRASituational characteristics during accident managementStructuring PIFs for human reliability analysisConclusionsAcknowledgementsReferences