Methods for Accident Investigation R O S S Reliability

April 9, 2019 | Author: Rony Jose | Category: Causality, Traffic Collision, Risk Management, Inductive Reasoning, Risk
Share Embed Donate


Short Description

Acc inv...

Description

Methods for accident invesgaon R O S S Reliability, Safety, and Security Studies at NTNU Dept. of roducon and !uality "ngineering #ddress$ %ising address$ Telephone$ &acsi'ile$ N()*+ Trondhei' S.. #ndersens vei - *) )/ -+ /0 11 *) )/ -+ ) ) T2T3" Methods for accident invesgaon #UT4OR Snorre S5let SUMM#R6 This report gives an overvie7 of so'e i'portant, recogni8ed, and co''only used 'ethods for invesgaon of 'a9or accidents. 2nvesgaon of 'a9or accidents usually caused by 'ulple, interrelated causal factors should be perfor'ed by a 'ul(disciplinary invesgaon tea', and supported by suitable, for'al 'ethods for accident invesgaon. "ach of the 'ethods has di:erent areas of applicaon and a set of 'ethods ought to be used in a co'prehensive accident invesgaon. The 'ethods dealt 7ith are li'ited to 'ethods used for in(depth analysis of 'a9or accidents. #R;42%" NTNU? =11=10 2S@N 0=())1A(0( D#T" =11=((1 S2BN#TUR" Marvin Rausand #B"SC#"ND. )- #"@? Method ......... -/ *.=.0 TR2OD................................................................................... -A *.=.+ #cci( 'ap.................................................................................. A * - D2S;USS2ON #ND ;ON;3US2ON................................................ A) -. D2S;USS2ON.................................................................................. A) -.= ;ON;3US2ON................................................................................ ) A R"&"R"N;"S ................................................................................... )/ Methods for accident invesgaon -  2ntroducon . 2ntroducon to accident invesgaon and deli'itaons of the report The accident invesgaon process consists of a 7ide range of acvies, and is described so'e7hat di:erent by di:erent authors. DO" >+++? divide the invesgaon process into three phasesK collecon of evidence and facts, analysis of these facts, and develop'ent of conclusions and develop'ent develop'ent of 9udg'ents of need and 7ring the report, see &igure . These are all overlapping phases and the 7hole process is iterave. So'e authors also include the i'ple'entaon and follo7(up of reco''endaons in the invesgaon phase >e.g., see &igure =? $  This approach is not li'ited to 'a9or accidents, but also include occupaonal accidents. accidents. ;ollecon of evidence and facts #nalysis of evidence and factsK Develop'ent of conclusions Develop'ent of 9udg'ents of needK ring the report Methods for accident invesgaon A . #ll reported incidents >accidents and near accidents? are invesgated i''ediately at the Hrst level by the supervisor and safety representave. =. # selecon of serious incidents, i.e. freGuently recurring types of incidents and incidents 7ith high loss potenal >actual or possible? are subseGuently invesgated by a proble'(solving group. /. On rare occasions, 7hen the actual or potenal loss is high, an accident invesgaon co''ission carries out the invesgaon. This co''ission has an independent status in relaon to the organisaons that are responsible for the occurrence. &igure =. #ccident invesgaon at three levels >Reason, ++)?. Organisaonal accidents are the co'paravely rare, but oen catastrophic, events that occur 7ithin co'ple, 'odern technologies such as nuclear po7er plants, co''ercial aviaon, petroche'ical industry, etc. Organisaonal accidents have 'ulple causes involving 'any people operang at di:erent levels of their respecve co'panies. @y contrast, individual accidents are accidents in 7hich a speciHc person or a group is oen both the agent and the vic' of the accident. Organisaonal accidents 2ndependant invesgaon co''ission or5 place roble'(solving group 2''ediate invesgaon by Hrst(line supervisor Reporng 2'ple'entaon of re'edial acons #ccidents Near accidents #ll events #ll events 2n eceponal cases &reGuent or severe events Methods for accident invesgaon ) are according to Reason >++)? a product of technological innovaons that have radically altered the relaonship bet7een syste's and their hu'an ele'ents. Ras'ussen >++)? proposes di:erent ris5 'anage'ent strategies for di:erent 5inds of accidents, see &igure /. The accident invesgaon 'ethods dealt 7ith in this report are li'ited to

'ethods used for evoluonary safety control, i.e. in(depth analysis of 'a9or accidents >ref. e.g., DO", ++)?, others focus on deter'ining factors >e.g., e.g., 4op5ins, =111?, acve failures and latent condions >e.g., Reason, ++)? or safety proble's >4endric5  @enner, +0)?. 4op5ins >=111? deHnes cause in the follo7ing 7ay$ Pone thing is said to be a cause of another if 7e can say but for the Hrst the second 7ould not have occurredQ. 3eplat >++)? epresses this in a 'ore for'al 7ay by saying that in general, the follo7ing type of deHnion of cause is accepted$ accepted$ Pto say that event  is the cause of event 6 is to say that the accidentQ period instead of on the 'igaon of the conseGuence of the accident. Methods for accident invesgaon + occurrence of  is a necessary condion to the producon of 6, in the circu'stances consideredQ. Such a deHnion i'plies that if any one of the causal path7ays idenHed are re'oved, the outco'e 7ould probably not have occurred. Using the ter' contribung factor 'ay be less for'al, if an event has not occurred, this 7ould necessarily not prevented the occurrence of the accident. =11? reco''ends avoiding the 7ord cause in accident invesgaons and rather tal5 about 7hat 'ight have prevented the accident. #ccident invesgators 'ay use di:erent fra'es for their analysis of accidents, but nevertheless the conclusions about 7hat happened, 7hy did it happen and 7hat 'ay be done in order to prevent future accidents 'ay be the sa'e. So'e deHnions are included in this chapter. These deHnions are 'eant as an introducon to the ter's. Several of the ter's are deHned in di:erent 7ays by di:erent authors. The deHnions are Guoted 7ithout any co''ents or discussions in this report in order to sho7 so'e of the specter. Therefore, these deHnions represent the authorsE

opinions. #ccident # seGuence of logically and chronologically related deviang events involving an incident that results in in9ury to personnel or da'age to the environ'ent or 'aterial assets. >DO", ++)? @arrier #nything used to control, prevent, or i'pede energy Fo7s. ;o''on types of barriers include eGuip'ent, ad'inistrave procedures and processes, supervisionC'anage'ent, 7arning devices, 5no7ledge and s5ills, and physical. @arriers 'ay be either control or safety. >DO", ++)? @arrier analysis #n analycal techniGue used to idenfy the energy sources and the failed or deHcient barriers and controls that contributed to an accident. >DO", ++)? Methods for accident invesgaon 1 ;ausal factor #n event or condion in the accident seGuence necessary and sucient to produce or contribute to the un7anted result. ;ausal factors fall into three categoriesK direct cause, contribung cause and root cause. >DO", ++)? ;ause of accident ;ontribung factor or root cause. >DO", ++)? ;ontribung factor More lasng ris5(increasing condion at the 7or5place related to design, organisaon or social syste'. >;;S, ++=? The causal factor>s? that, if corrected, 7ould prevent recurrence of the accident. >DO", ++)? Most basic cause of an accidentCincident, i.e. a lac5 of adeGuate 'anage'ent control resulng in deviaons and contribung factors. >bla'e?  "valuate the Gueson of guilt in order to assess the liability for co'pensaon >pay? #s 7e see, there 'ay be di:erent purposes in 7hich iniate accident invesgaons. The di:erent purposes 7ill not be discussed any'ore in this report. =.* Responsibility for accident invesgaon ho should be responsible for perfor'ing accident invesgaons and ho7 thoroughly should the accident be invesgatedI The history of accident invesgaon in the past decades sho7s a trend to go further and further bac5 in the analysis, i.e., fro' being sasHed 7ith idenfying hu'an errors by front(personnel or technical failures to idenfy 7ea5nesses in the govern'ental policies as root causes. 2n order to 5no7 7hen 7e should stop our invesgaon, 7e need 7hat Ras'ussen >++1? called stop(rules. Reason >++)? suggests that 7e should stop 7hen the causes idenHed are no longer controllable. The stopping rule suggested by Reason >++)?, leads to di:erent stopping points for di:erent pares. ;o'panies should trace causes bac5 to failures in their o7n 'anage'ent syste's and develop ris5reducing 'easures that they have authority to i'ple'ent. Supervisory authories >e.g., The Nor7egian etroleu' Directorate?, appointed govern'ental co''issions of inGuiries >e.g., the Sleipnerco''ission, and the Vsta(co''ission? or per'anent invesgaon boards >e.g., The Nor7egian #ircra #ccident 2nvesgaon @oard? Methods for accident invesgaon A should in addion focus on regulatory syste's and as5 7hether 7ea5nesses in these syste's contributed to the accident. The police and the prosecung authority are responsible for evaluang the basis for potenal cri'inal prosecuon, 7hile the court of 9usce is responsible for passing sentence on a person or a co'pany. The liability for co'pensaon is 7ithin the insurance co'paniesE and the la7yerEs range of responsibility. =.- ;riteria for accident invesgaons hat is a PgoodQ accident invesgaonI This Gueson is dicult to ans7er in a si'ple 7ay, because the ans7er depends on the purpose of the invesgaon. Nevertheless, 2 have included ten funda'ental criteria for accident invesgaons stated by 4endric5  @enner >+0)?. Three criteria are related to ob9ecves and

purposes of the accident invesgaon, four to invesgave procedures, and three to the outputs fro' the invesgaon and its usefulness. ;riteria related to ob9ecves and purposes  Realisc The invesgaon should result in a realisc descripon of the events that have actually occurred.  Non( causal #n invesgaon should be conducted in a non(causal fra'e7or5 and result in an ob9ecve descripon of the accident process events. #Wribuon of cause or fault can only be considered separate fro', and aer the understanding of the accident process is co'pleted to sasfy this criterion.  ;onsistent The invesgaon perfor'ance fro' accident to accident and a'ong invesgaons of a single accident to di:erent invesgators should be consistent. Only consistency bet7een results of di:erent invesgaons enables co'parison bet7een the'. Methods for accident invesgaon ) ;riteria related to invesgaon procedures  Disciplining #n invesgaon process should provide an orderly, syste'ac fra'e7or5 and set of procedures to discipline the invesgatorsE tas5s in order to focus their e:orts on i'portant and necessary tas5s and avoid duplicave or irrelevant tas5s.  &unconal #n invesgaon process should be funconal in order to 'a5e the 9ob ecient, e.g. by helping the invesgator to deter'ine 7hich events 7ere part of the accident process as 7ell as those events that 7ere unrelated.  DeHnive #n invesgaon process should provide criteria to idenfy and deHne the data that is needed to describe 7hat happened.  ;o'prehensive #n invesgaon process should be co'prehensive so there is no confusion about 7hat happened, no unsuspected gaps or holes in the eplanaon, and no conFict of understanding a'ong those 7ho read the report. ;riteria related to output and usefulness  Direct The invesgaon process should provide results that do not reGuire collecon of 'ore data before the needed controls can be idenHed and changes 'ade.  Understandable The output should be readily understandable.  Sasfying The results should be sasfying for those 7ho inialised the invesgaon and other individuals that de'and results fro' the invesgaons. So'e of these criteria are debatable. &or instance 7ill the second criterion related to causality be disputable. 2nvesgators using the causal(seGuence accident 'odel 7ill in principle focus on causes during their invesgaon process. #lso the last criterion related to sasfacon 'ight be discussed. 2'agine an invesgaon inialised by the top 'anage'ent in a co'pany. 2f the top 'anage'ent is cricised Methods for accident invesgaon 0 in the accident report, they are not necessarily sasHed 7ith the results, but nevertheless it 'ay be a PgoodQ invesgaon. Methods for accident invesgaon + / The accident invesgaon process &igure - sho7s the detailed accident invesgaon process as described by DO" >+++?. #s sho7n in the Hgure, the process starts i''ediately 7hen an accident occurs, and the 7or5 is not Hnished before the Hnal report is accepted by the appoinng ocial. This report focuses on the process of analysing evidence to deter'ine and evaluate causal factors >see chapter *?, but Hrst a fe7 co''ents to the other 'ain phases. &igure -. DO"Es process for accident invesgaon >DO", +++?. @oard acvites #ccident occurs Develop conclusions and deter'ine 9udg'ents of need "valuate causal factors 2ntegrate, organise, and analyse evidence to deter'ine causal factors ;ollect, preserve, and verify evidence @oard chairperson ta5es responsibility for accident scene @oard arrives at accident scene #ppoinng ocial Selects @oard chairperson and 'e'bers Readiness tea' responds Secures scene Ta5es 7itness state'ents reserves evidence 2nial reporng and categorisaon ;onduct reGuire'ents veriHcaon analysis repare dra report @oard 'e'bers Hnalise dra report #ppoinng ocial accepts report Site organisaons conduct fractual accurace revie7 @oard chairperson conducts closeout brieHng Methods for accident invesgaon =1 /. ;ollecng evidence and facts ;ollecng data is a crical part of  the invesgaon. Three 5ey types of evidence are collected during the invesgaon process$  4u'an or testa'entary evidence 4u'an or testa'entary evidence includes 7itness state'ents and observaons.  hysical evidence hysical evidence is 'aWer related to the accident >e.g. eGuip'ent, parts, debris,

hard7are, and other physical ite's?.  Docu'entary evidence Docu'entary evidence includes paper and electronic infor'aon, such as records, reports, procedures, and docu'entaon. The 'a9or steps in gathering evidence are collecng hu'an, physical and docu'entary evidence, ea'ining organisaonal concerns, 'anage'ent syste's, and line 'anage'ent oversight and at last preserving and controlling the collected evidence. ;ollecng evidence can be a lengthy, 'e(consu'ing, and piece'eal process. itnesses 'ay provide s5etchy or conFicng accounts of the accident. hysical evidence 'ay be badly da'aged or co'pletely destroyed, Docu'entary evidence 'ay be 'ini'al or dicult to access. Thorough invesgaon reGuires that board 'e'bers are diligent in pursuing evidence and adeGuately eplore leads, lines of inGuiry, and potenal causal factors unl they gain a suciently co'plete understanding of the accident. This topic 7ill not be discussed any'ore in this report, but for those interested in the topic are the follo7ing references usefulK DO" >+++?, ;;S >++=? and 2ngstad >+00?. Methods for accident invesgaon = /.= #nalysis of evidence and facts #nalysis of evidence and facts is the process of deter'ining causal factors, idenfy latent condions or contribung factors >or 7hatever you 7ant to call it? and see5s to ans7er the follo7ing t7o Guesons$  hat happened 7here and 7henI  hy did it happenI DO" >+++? describes three types of causal factors$ . Direct cause =. ;ontribung causes /. Root causes # direct cause is an i''ediate event or condion that caused the accident >DO", ++)?. # contribung cause is an event or condion that together 7ith other causes increase the li5elihood of an accident but 7hich individually did not cause the accident >DO", ++)?. # root cause is the causal factor>s? that, if corrected, 7ould prevent recurrence of the accident >DO", ++)?. There are di:erent opinions of the concept of causality of accidents, see co''ents in secon .=., but this topic 7ill not be discussed any further here. ;;S >++=? lists three analycal approaches by 7hich conclusions can be reached about an accident$  Deducve approach  2nducve approach.  Morphological approach 2n addion, there eists di:erent concepts for accident invesgaon not as co'prehensive as these syste'(oriented techniGues. These are categori8ed as non(syste'(oriented techniGues. The deducve approach involves reasoning fro' the general to the speciHc. 2n the deducve analysis, it is postulated that a syste' or process has failed in a certain 7ay. Net an aWe'pt is 'ade to deter'ine 7hat 'odes of syste', co'ponent, operator and organisaon behaviour contribute to the failure. The 7hole accident Methods for accident invesgaon == invesgaon process is a typical ea'ple of a deducve reasoning. &ault tree analysis is also an ea'ple of a deducve techniGue. The inducve approach involves reasoning fro' individual cases to a general conclusion. #n inducve analysis is perfor'ed by postulang that a parcular fault or iniang event has occurred. 2t is then deter'ined 7hat the e:ects of the fault or iniang event are on the syste' operaon. ;o'pared 7ith the deducve approach, the inducve approach is an Povervie7Q 'ethod. #s such it bring an overall structure to the invesgave process. To probe the details of the causal factors, control and barrier funcon, it is oen necessary to apply deducve analysis. "a'ples of inducve techniGues are failure 'ode and e:ects analysis >&M";#?, 4#XOEs and event tree analysis. The 'orphological approach to analycal incident invesgaon is based on the structure of the syste' being studied. The 'orphological approach focuses directly on potenally ha8ardous ele'ents >for ea'ple operaon, situaons?. The ai' is to concentrate on the factors having the 'ost signiHcant inFuence on safety. hen perfor'ing a 'orphological analysis, the analyst is pri'arily applying his or her past eperience of incident invesgaon. Rather than loo5ing at all possible deviaons 7ith and 7ithout a potenal safety i'pact, the invesgaon focuses on 5no7n ha8ard sources. Typically, the 'orphological approach is an adaptaon of deducve or inducve approaches, but 7ith its o7n guidelines. S2NT"& has developed a useful Hve(step 'odel for invesgaon of causes of accidents. The 'odel is illustrated in &igure A. Step 

is idenHcaon of the event seGuences 9ust before the accident. Step = is idenHcaon of deviaons and failures inFuencing the event seGuence that led to the accident. This includes deviaons fro' eisng procedures, deviaons fro' co''on pracce, technical failures and hu'an failures. Step / is idenHcaon of 7ea5nesses and defects 7ith the 'anage'ent syste's. The ob9ecve is to detect possible causes of the deviaons or failures idenHed in Step =. Step * is idenHcaon of 7ea5nesses and defects related to the top 'anage'ent of the co'pany, because it is their responsibility to establish the necessary 'anage'ent syste's and ensure that the syste's are co'plied 7ith. Step - is idenHcaon of potenal Methods for accident invesgaon =/ deHciencies related to the public safety fra'e7or5, i.e. 'ar5ed condions, la7s and regulaons. &igure A. S2NT"&Es 'odel for analysis of accident causes >#rbeids'il9Ysenteret, =11?. Di:erent 'ethods for analysis of evidence and facts are further discussed in chapter *. DeHciencies related to the public safety fra'e7or5 Z "cono'y Z 3abour Z 3a7s and regulaons etc. "vent seGuence Z Decisions Z #cons Z O'issions Deviaons and failures inFuencing the event seGuence Z rocedures not follo7ed Z Technical failures Z 4u'an failures ea5nesses and defects 7ith the 'anage'ent syste's Z 3ac5 of or inadeGuate procedures Z 3ac5 of i'ple'entaon Z 2nsucient trainingCeducaon Z 2nsucient follo7(up ea5nesses and defects related to the top 'anage'ent Z olicy Z Organisaon and responsibilites Z 2nFuence on aJtudes Z &ollo7(up by 'anage'ent Undesirable event 3oss C in9uries on Z ersonnel Z roperes Z "nviron'ent #nalysis of causes #nalysis of conseGuences Step  Step - Step * Step / Step = #nalysis of organisaon ST"analysis Methods for accident invesgaon =* /./ Reco''endaons and reporng One of the 'ain ob9ecves of perfor'ing accidents invesgaons is to idenfy reco''endaons that 'ay prevent the occurrence of future accidents. This topic 7ill not be discussed any further, but the reco''endaons should be based on the analysis of evidence and facts in order to prevent that the revealed direct and root causes 'ight lead to future accidents. #t the co'pany level the reco''ended ris5 reducing 'easures 'ight be focused on technical, hu'an, operaonal andCor organisaonal factors. Oen, it is even 'ore i'portant to focus aWenon to7ards changes in the higher levels in &igure *, e.g., by changing the regulaons or the authoritave supervisory pracce. # useful p is to be open('inded in the search for ris5 reducing 'easures and not to be narro7 in this part of the 7or5. 4endric5 and @enner >+0)? says that t7o thoughts should be 5ept in 'ind regarding accident reports$  2nvesgaons are re'e'bered trough their reports  The best invesgaon 7ill be 7asted by a poor report. Methods for accident invesgaon =- * Methods for accident invesgaons # nu'ber of 'ethods for accident invesgaon have been developed, 7ith their o7n strengths and 7ea5nesses. So'e 'ethods of great i'portance are selected for further ea'inaon in this chapter. The selecon of 'ethods for further descripon is not based on any scienHc selecon criteria. @ut the 'ethods are 7idely used in pracce, 7ell ac5no7ledged, 7ell described in the literature* and so'e 'ethods that are relavely recently developed. 2n order to sho7 the span in di:erent accident invesgaon 'ethods, Table  sho7s an oversight over 'ethods described by DO" >+++? and Table = sho7s an oversight described by ;;S >++=?. So'e of the 'ethods in the tables are overlapping, 7hile so'e are di:erent. Table . #ccident invesgaon analycal techniGues presented in DO" >+++?. ;ore #nalycal TechniGues "vents and ;ausal &actors ;harng and #nalysis @arrier #nalysis ;hange #nalysis Root ;ause #nalysis ;o'ple #nalycal TechniGues &or co'ple accidents 7ith 'ulple syste' failures, there 'ay in addion be need of analycal techniGues li5e analyc tree analysis, e.g. &ault Tree #nalysis MORT >Manage'ent Oversight and Ris5 Tree? "T >ro9ect "valuaon Tree #nalysis? SpeciHc #nalycal TechniGues 4u'an &actors #nalysis 2ntegrated #ccident "vent Matri &ailure Modes and ":ects #nalysis So7are 4a8ards #nalysis ;o''on ;ause &ailure #nalysis Snea5 ;ircuit #nalysis )=(4our roHle Materials and Structural #nalysis ScienHc Modelling

>e.g., for incidents involving cricality and at'ospheric despersion? * So'e 'ethods are co''ercialised and therefore li'ited described in the public available literature. Methods for accident invesgaon =A Table =. #ccident invesgaons 'ethods described by ;;S >++=?. 2nvesgaon 'ethod #ccident #nato'y 'ethod >##M? #con "rror #nalysis >#"#? #ccident "voluon and @arrier #nalysis >#"@? ;hange "valuaonC#nalysis ;ause(":ect 3ogic Diagra' >;"3D? ;ausal Tree Method >;TM? &ault Tree #nalysis >&T#? 4a8ard and Operability Study >4#XO? 4u'an erfor'ance "nhance'ent Syste' >4"S? 4u'an Reliability #nalysis "vent Tree >4R#("T? Mulple(;ause, Syste's(oriented 2ncident 2nvesgaon >M;SO22? Mullinear "vents SeGuencing >M"S? Manage'ent Oversight Ris5 Tree >MORT? Syste'ac ;ause #nalysis TechniGue >S;#T? SeGuenally Ti'ed "vents loJng >ST"? TapRoot[ 2ncident 2nvesgaon Syste' TechniGue of Operaons Revie7 >TOR? or5 Safety #nalysis  roprietary techniGues that reGuires a license agree'ent. These t7o tables list 'ore than =1 di:erent 'ethods, but do not include a co'plete list of 'ethods. Other 'ethods are described else7here in the literature. Since DO"Es or5boo5 ;onducng #ccident 2nvesgaon >DO", +++? is a co'prehensive and 7ell( 7riWen handboo5, the descripon of accident invesgaon 'ethods starts 7ith DO"Es core analycal techniGues in secon *.. Their core analycal techniGues are$  "vents and ;ausal &actors ;harng and #nalysis  @arrier #nalysis  ;hange #nalysis  Root ;ause #nalysis &urther, so'e other 'ethods are described in secon *.=$  &ault tree analysis  "vent tree analysis  MORT >Manage'ent Oversight and Ris5 Tree? Methods for accident invesgaon =)  S;#T >Syste'ac ;ause #nalysis TechniGue?  ST" >SeGuenal Ti'ed "vents loJng?  MTO(analysis  #"@ Method  TR2OD(Delta  #cci(Map The four last 'ethods are neither listed in Table  nor Table =, but are co''only used 'ethods in di:erent industries in several "uropean countries. The readers should be a7are of that this chapter is purely descripve. #ny co''ents or assess'ents of the 'ethods are 'ade in chapter -. *. DO"Es core analycal techniGues- *.. "vents and causal factors charng >";&;? "vents and causal factors charng is a graphical display of the accidentEs chronology and is used pri'arily for co'piling and organising evidence to portray the seGuence of the accidentEs events. The events and causal factor chart is easy to develop and provides a clear depicon of the data. DO", +++?A . A Si'ilar to M"S in Table =. ;ondion #ccident event ;ondion Secondary event  "vent  Secondary event = "vent = "vent / "vent * Secondary events seGuence ri'ary events seGuence ;ondion ;ondion ;ondion "vents #ccidents ;ondions ;onnector Transfer bet7een lines 3T# 3ess than adeGuate >9udg'ent? ( #re acve >e.g. ]crane stri5es building]? ( Should be stated using one noun and one acve verb ( Should be GuanHed as 'uch as possible and 7here applicable ( Should indicate the date and 'e, 7hen they are 5no7n ( Should be derived fro' the event or events and conditons i''ediately preceding it ( #re passive >e.g. ]fog in the area]? ( Describe states or circu'stances rather than occurrences or events ( #s praccal, should be GuanHed ( Should indicate date and 'e if praccalCapplicable ( #re associated 7ith the corresponding event "nco'passes the 'ain events of the accident and those that for' the 'ain events line of the chart "nco'passes the events that are secondary or contribung events and those that for' the secondary line of the chart Secondary event seGuence ri'ary event seGuence ;ondions Sy'bols "vents resu'pve events resu'pve condions or assu'pons Methods for accident invesgaon /1 *..= @arrier analysis @arrier analysis is used to idenfy ha8ards associated 7ith an accident and the barriers that should have been in place to prevent it. # barrier is any 'eans used to control, prevent, or i'pede the ha8ard fro' reaching the target. @arrier analysis addresses$  @arriers that 7ere in place and ho7 they perfor'ed  @arriers that 7ere in place but not used  @arriers that 7ere not in place but 7ere reGuired  The barrier>s? that, if present or strengthened, 7ould prevent the sa'e or si'ilar accidents fro' occurring in the future. &igure + sho7s types of barriers that 'ay be in place to protect 7or5ers fro' ha8ards. &igure +. "a'ples on barriers to protect 7or5ers fro' ha8ards >DO", +++?) hysical barriers are usually easy to idenfy, but 'anage'ent syste' barriers 'ay be less obvious >e.g. eposure li'its?. The invesgator 'ust understand each barrierEs intended funcon and locaon, and ho7 it failed to prevent the accident. There eists di:erent 7ays in ) There eists di:erent barrier 'odels for prevenon of accidents based on the defence(in(depth principle in di:erent industries, see. e.g. =111? for prevenon of Hres and eplosions in hydrocarbon processing plants and 2NS#B(= for basic safety principles for nuclear po7er plants. Types of barriers Manage'ent barriers ( 4a8ard analysis ( hat 7as the barrierEs purposeI as the barrier in place or not in placeI Did the barrier failI as the barrier used if it 7as in placeI? Record in colu'n t7o. Step * 2denfy and consider probable causes of the barrier failure. Record in colu'n three. Step - "valuate the conseGuences of the failure in this accident. Record in colu'n four. Methods for accident invesgaon /= Table /. @arrier analysis 7or5sheet. 4a8ard$ /.= 5% electrical cable Target$ #cng pipeHWer hat 7ere the barriersI 4o7 did each barrier perfor'I hy did the barrier failI 4o7 did the barrier a:ect the accidentI "ngineering dra7ings Dra7ings 7ere inco'plete and did not idenfy electrical cable at su'p locaon "ngineering dra7ings and construcon speciHcaons 7ere not procured Dra7ings used 7ere preli'inary No as(built dra7ings 7ere used to idenfy locaon of ulity lines "istence of electrical cable un5no7n 2ndoor ecavaon per'it 2ndoor ecavaon per'it 7as not obtained ipeHWers and ulity specialist 7ere una7are of indoor ecavaon per'it reGuire'ents Opportunity to idenfy eistence of cable 'issed *../ ;hange analysis ;hange is anything that disturbs the PbalanceQ of a syste' operang as planned. ;hange is oen the source of deviaons in syste' operaons. ;hange analysis ea'ines planned or unplanned changes that caused undesired outco'es. 2n an accident invesgaon, this techniGue is used to ea'ine an accident by analysing the di:erence bet7een 7hat has occurred before or 7as epected and the actual seGuence of events. The invesgator perfor'ing the change analysis idenHes speciHc di:erences bet7een the accident\free situaon and the accident scenario. These di:erences are evaluated to deter'ine 7hether the di:erences caused or contributed to the accident. The change analysis process is described in &igure . hen conducng a change analysis, invesgators idenfy changes as 7ell as the results of those changes. The disncon is i'portant, because idenfying only the results of change 'ay not pro'pt invesgators to Methods for accident invesgaon // idenfy all causal factors of an accident. hen conducng a change analysis, it is i'portant to have a baseline situaon that the accident seGuence 'ay be co'pared to. &igure . The change analysis process. >DO", +++? Table * sho7s a si'ple change analysis 7or5sheet. The invesgators should Hrst categorise the changes according to the Guesons sho7n in the le colu'n of the 7or5sheet, i.e., deter'ine if the change pertained to, for ea'ple, a di:erence in$  hat events, condions, acvies, or eGuip'ent 7ere present in the accident situaon that 7ere not present in the baseline >accident(free, prior, or ideal? situaon >or vice versa?  hen an event or condion occurred or 7as detected in the accident situaon versus the baseline situaon  here an event or condion occurred in the accident situaon versus 7here an event or condion occurred in the baseline situaon  ho 7as involved in planning, revie7ing, authorising, perfor'ing, and supervising the 7or5 acvity in the accident versus the accident(free situaon.  4o7 the 7or5 7as 'anaged and controlled in the accident versus the accident(free situaon. To co'plete the re'ainder of the 7or5sheet, Hrst describe each event or condion of interest in the second colu'n. Then describe the related event or condion that occurred >or should have occurred? in the baseline situaon in the third colu'n. The di:erence bet7een the event and condions in the accident and the baseline situaons should Describe accident situaon Describe co'parable accident(free situaon 2nput results into events and causal factors chart #nalyse di:erences for e:ect on accident 2denfy di:erences ;o'pare Methods for accident invesgaon /* be brieFy described in the fourth colu'n. 2n the last colu'n, discuss the e:ect that each change had on the accident. The di:erences or changes idenHed can generally be described as causal

factors and should be noted on the events and causal factors chart and used in the root cause analysis. # potenal 7ea5ness of change analysis is that it does not consider the co'pounding e:ects of incre'ental change >for ea'ple, a change that 7as instuted several years earlier coupled 7ith a 'ore recent change?. To overco'e this 7ea5ness, invesgators 'ay choose 'ore than one baseline situaon against 7hich to co'pare the accident scenario. Table *. # si'ple change analysis 7or5sheet. >DO", +++? &actors #ccident situaon rior, ideal, or acciden^ree situaon Di:erence "valuaon of e:ect hat ;ondions Occurrences #cvies "Guip'ent hen Occurred 2denHed &acility status Schedule here hysical locaon "nviron'ental condions ho Sta: involved Training !ualiHcaon Supervision 4o7 ;ontrol chain 4a8ard analysis Monitoring Other Methods for accident invesgaon /- *..* "vents and causal factors analysis The events and causal factors chart 'ay also be used to deter'ine the causal factors of an accident, as illustrated in &igure =. This process is an i'portant Hrst step in later deter'ining the root causes of an accident. "vents and causal factors analysis reGuires deducve reasoning to deter'ine 7hich events andCor condions that contributed to the accident. &igure =. "vents and causal factors analysis. >DO", +++? @efore starng to analyse the events and condions noted on the chart, an invesgator 'ust Hrst ensure that the chart contains adeGuate detail. "a'ine the Hrst event that i''ediately precedes the accident. "valuate its signiHcance in the accident seGuence by as5ing$ P2f this event had not occurred, 7ould the accident have occurredIQ 2f the ans7er is yes, then the event is not signiHcant. roceed to the net event in the chart, 7or5ing bac57ards fro' the accident. 2f the ans7er is no, then deter'ine 7hether the event represented nor'al acvies 7ith the epected conseGuences. 2f the event 7as intended and had the epected outco'es, then it is not signiHcant. 4o7ever, if the event deviated fro' 7hat 7as intended or had un7anted conseGuences, then it is a signiHcant event. ;ondion ;ausal factor ;ausal factor ;ondion ;ondion "vent "vent "vent "vent 4o7 did the condions originateI hy did the syste' allo7 the condions to eistI hy did this event happenI #s5 Guesons to deter'ine causal factors >7hy, ho7, 7hat, and 7ho? "vent chain Methods for accident invesgaon /A ;arefully ea'ine the events and condions associated 7ith each signiHcant event by as5ing a series of Guesons about this event chain, such as$  hy did this event happenI  hat events and condions led to the occurrence of the eventI  hat 7ent 7rong that allo7ed the event to occurI  hy did these condions eistI  4o7 did these condions originateI  ho had the responsibility for the condionsI  #re there any relaonships bet7een 7hat 7ent 7rong in this event chain and other events or condions in the accident seGuenceI  2s the signiHcant event lin5ed to other events or condions that 'ay indicate a 'ore general or larger deHciencyI The signiHcant events, and the events and condions that allo7ed the signiHcant events to occur, are the accidentEs causal factors. *..- Root cause analysis Root cause analysis is any analysis that idenHes underlying deHciencies in a safety 'anage'ent syste' that, if corrected, 7ould prevent the sa'e and si'ilar accidents fro' occurring. Root cause analysis is a syste'ac process that uses the facts and results fro' the core analyc techniGues to deter'ine the 'ost i'portant reasons for the accident. hile the core analyc techniGues should provide ans7ers to Guesons regarding 7hat, 7hen, 7here, 7ho, and ho7, root cause analysis should resolve the Gueson 7hy. Root cause analysis reGuires a certain a'ount of 9udg'ent. # rather ehausve list of causal factors 'ust be developed prior to the applicaon of root cause analysis to ensure that Hnal root causes are accurate and co'prehensive. One 'ethod for root cause analysis described by DO" is T2"R diagra''ing. T2"R(diagra''ing is used to idenfy both the root causes of an accident and the level of line 'anage'ent that has the responsibility and authority to correct the accidentEs causal factors. The invesgators use T2"R(diagra's to hierarchically categorise the causal factors derived fro' the events and causal factors analysis. Methods for accident invesgaon /)

3in5ages a'ong causal factors are then idenHed and possible root causes are developed. # di:erent diagra' is developed for each organisaon responsible for the 7or5 acvies associated 7ith the accident. The causal factors idenHed in the events and causal factors chart are input to the T2"R( diagra's. #ssess 7here each causal factor belong in the T2"R(diagra'. #er arranging all the causal factors, ea'ine the causal factors to deter'ine 7hether there is lin5age bet7een t7o or 'ore of the'. "valuate each of the causal factors state'ents if they are root causes of the accident. There 'ay be 'ore than one root cause of a parcular accident. &igure / sho7s an ea'ple on a T2"R(diagra'. &igure /. 2denfying the lin5ages to the root causes fro' a T2"R(diagra'. *.= Other accident invesgaon 'ethods *.=. &ault tree analysis0 &ault tree analysis is a 'ethod for deter'ining the causes of an accident >or top event?. The fault tree is a graphic 'odel that displays the various co'binaons of nor'al events, eGuip'ent failures, hu'an errors, and environ'ental factors that can result in an accident. #n ea'ple of a fault tree is sho7n in &igure *. 0 The descripon is based on 4Yyland  Rausand, ++*. Tier ;ausal &actors Tier -$ Senior 'anage'ent Tier $ or5er acons Tier =$ Supervision Tier /$ 3o7er 'anage'ent Tier *$ Middle 'anage'ent Tier 1$ Direct cause Root causes >oponal colu'n? Root cause _  Root cause _ / Root cause _ = Methods for accident invesgaon /0 &igure *. 2llustraon of a fault tree >ea'ple fro' the Vsta(accident?. # fault tree analysis 'ay be Gualitave, Guantave, or both. ossible results fro' the analysis 'ay be a lisng of the possible co'binaons of environ'ental factors, hu'an errors, nor'al events and co'ponent failures that 'ay result in a crical event in the syste' and the probability that the crical event 7ill occur during a speciHed 'e interval. The strengths of the fault tree, as a Gualitave tool is its ability to brea5 do7n an accident into root causes. The undesired event appears as the top event. This event is lin5ed to the basic failure events by logic gats and event state'ents. # gate sy'bol can have one or 'ore inputs, but only one output. # su''ary of co''on fault tree sy'bols is given in &igure -. 4Yyland and Rausand >++*? give a 'ore detailed descripon of fault tree analysis. Malfuncon of the signalling syste' 4u'an error >engine driver? 3ine secon already ]occupied] by another train SabotageC act of terros "ngine failure >runa7ay train? Or Or No signal Breen signal >green Fash? Methods for accident invesgaon /+ &igure -. &ault tree sy'bols. *.=.= "vent tree analysis+ #n event tree is used to analyse event seGuences follo7ing aer an iniang event. The event seGuence is inFuenced by either success or failure of nu'erous barriers or safety funconsCsyste's. The event seGuence leads to a set of possible conseGuences. The conseGuences 'ay be considered as acceptable or unacceptable. The event seGuence + The descripon is based on %ille'eur, ++. # #nd " " "= / # Or " " "= / The OR(gate indicates that the output event # occurs if any of the input events "i occur. The #ND(gate indicates that the output event # occurs 7hen all the input events "i occur si'ultaneously. @asic event Undeveloped event ;o''ent rectangle The basic event represents a basic eGuip'ent failure that reGuires no further develop'ent of failure causes The undeveloped event represents an event that is not ea'ined further because infor'aon is unavailable or because its conseGuences is insigniHcant The co''ent rectangle is for supple'entary infor'aon The transfer(out sy'bol indicates that the fault tree is developed further at the occurrence of the corresponding Transfer(in sy'bol Transfer(out Transfer(in 3ogic gates 2nput events Descripon of state Transfer sy'bols #ND(gate OR(gate Sy'bol Descripon Methods for accident invesgaon *1 is illustrated graphically 7here each safety syste' is 'odelled for t7o states, operaon and failure. &igure A illustrates an event tree of the situaon on RYrosbanen 9ust before the Vsta(accident. This event tree reveals the lac5 of reliable safety barriers in order to prevent train collision at RYrosbanen at that 'e. #n event tree analysis is pri'arily a proacve ris5 analysis 'ethod used to idenfy possible event seGuences. The event tree 'ay be used to idenfy and illustrate event seGuences and also to obtain a

Gualitave and Guantave representaon and assess'ent. 2n an accident invesgaon 7e 'ay illustrate the accident path as one of the possible event seGuences. This is illustrated 7ith the thic5 line in &igure A. &igure A. Si'pliHed event tree analysis of the ris5 at RYrosbanen 9ust before the Vsta( accident. *.=./ MORT1 MORT provides a syste'ac 'ethod >analyc tree? for planning, organising, and conducon a co'prehensive accident invesgaon. Through MORT analysis, invesgators idenfy deHciencies in speciHc 1 The descripon is based on ohnson .B., +01. T7o trains at the sa'e secon of the line #T; >#uto'ac Train ;ontrol? The rail trac controller detects the ha8ardous situaon Train drivers stop the train The rail trac controller alerts about the ha8ard 6es 6es 6es 6es No No No No ;ollision ;ollison avoided ;ollision ;ollision ;ollison avoided Methods for accident invesgaon * control factors and in 'anage'ent syste' factors. These factors are evaluated and analysed to idenfy the causal factors of the accident. @asically, MORT is a graphical chec5list in 7hich contains generic Guesons that invesgators aWe'pt to ans7er using available factual data. This enables invesgators to focus on potenal 5ey causal factors. The upper levels of the MORT diagra' are sho7n in &igure ). MORT reGuires etensive training to e:ecvely perfor' an in(depth analysis of co'ple accidents involving 'ulple syste's. The Hrst step of the process is to select the MORT chart for the safety progra' area of interest. The invesgators 7or5 their 7ay do7n through the tree, level by level. "vents should be coded in a speciHc colour relave to the signiHcance of the accident. #n event that is deHcient, or 3ess Than #deGuate >3T#? in MORT ter'inology is 'ar5ed red. The sy'bol is circled if suspect or coded in red if conHr'ed. #n event that is sasfactory is 'ar5ed green in the sa'e 'anner. Un5no7ns are 'ar5ed in blue, being circled inially and coloured if sucient data do not beco'e available, and an assu'pon 'ust be 'ade to connue or conclude the analysis. hen the appropriate seg'ents of the tree have been co'pleted, the path of cause and e:ect >fro' lac5 of control by 'anage'ent, to basic causes, contributory causes, and root causes? can easily be traced bac5 through the tree. The tree highlights Guite clearly 7here controls and correcve acons are needed and can be e:ecve in prevenng recurrence of the accident. Methods for accident invesgaon *= &igure ). The upper levels of the MORT(tree. "T >ro9ect "valuaon Tree? and SMORT >Safety Manage'ent and Organisaons Revie7 TechniGue? are both 'ethods based on MORT but si'pliHed and easier to use. "T and SMORT 7ill not be described further. "T is described by DO" >+++? and SMORT by +0)?. *.=.* Syste'ac ;ause #nalysis TechniGue >S;#T? The 2nternaonal 3oss ;ontrol 2nstute >23;2? developed S;#T for the support of occupaonal incident invesgaon. The 23;2 3oss ;ausaon Model is the fra'e7or5 for the S;#T syste' >see &igure 0?.  The descripon of S;#T is based on ;;S >++=? and the descripon of the 23;2('odel is based on @ird  Ber'ain >+0-?. 2n9uries, da'age, other costs, perfor'ance lost or degraded &uture undesired events  Or 2'ple'entaon 3T# #ccident #'elioraon 3T# Manage'ent syste' factos 3T# SpeciHc controls factors 3T# Oversights and o'issions #ssu'ed ris5s Ris5 assess'ent syste' 3T# Ris5  Ris5 = Ris5 / Ris5 n #nd Or Or olicy 3T# T SCM S M S# S#= M# M#/ M#= Or hat happenedI hyI # @ ; D R # Dra7ing brea5. Transfer to secon of tree indicated by sy'bol idenHcaon leWer(nu'ber Methods for accident invesgaon */ &igure 0. The 23;2 3oss ;ausaon Model >@ird and Ber'ain, +0-?. The result of an accident is loss, e.g. har' to people, properes, products or the environ'ent. The incident >the contact bet7een the source of energy and the Pvic'Q? is the event that precedes the loss. The i''ediate causes of an accident are the circu'stances that i''ediately precede the contact. They usually can be seen or sensed. &reGuently they are called unsafe acts or unsafe condions, but in the 23;2('odel the ter's substandard acts >or pracces? and substandard condions are used. Substandard acts and condions are listed in &igure +. &igure +. Substandard acts and condions in the 23;2('odel. @asic causes are the diseases or real causes

behind the sy'pto's, the reasons 7hy the substandard acts and condions occurred. @asic causes help eplain 7hy people perfor' substandard pracces and 3ac5 of control 2ncident 3oss 2''ediate causes @asic causes 2nadeGuate$ rogra' rogra' standards ;o'pliance to standards ersonal factors ob factors eople roperty roduct "nviron'ent Service ;ontact 7ith energy, substance or people Substandard acts Substandard condions Substandard praccesCacts Substandard condions . Operang eGuip'ent 7ithout authority =. &ailure to 7arn /. &ailure to secure *. Operang at i'proper speed -. Ma5ing safety devices inoperable A. Re'oving safety devices ). Using defecve eGuip'ent 0. Using eGuip'ent i'properly +. &ailing to use personal protecve eGuip'ent 1. 2'proper loading . 2'proper place'ent =. 2'proper liing /. 2'proper posion for tas5 *. Servicing eGuip'net in operaon -. 4orseplay A. Under inFuence of alcoholCdrugs . 2nadeGuate guards or barriers =. 2nadeGuate or i'proper protecve eGuip'ent /. Defecve tools, eGuip'ent or 'aterials *. ;ongeson or restricted acon -. 2nadeGuate 7arning syste' A. &ire and eplosion ha8ards ). oor house5eeping, disorderly 7or5place 0. 4a8ardous environ'ental condions +. Noise eposures 1. Radiaon eposures . 4igh or lo7 te'perature eposures =. 2nadeGuate or ecessive illu'inaon /. 2nadeGuate venlaon Methods for accident invesgaon ** 7hy substandard condions eists. #n overvie7 of personal and 9ob factors are given in &igure =1. &igure =1. ersonal and 9ob factors in the 23;2('odel. There are three reasons for lac5 of control$ . 2nadeGuate progra' =. 2nadeGuate progra' standards and /. 2nadeGuate co'pliance 7ith standards &igure = sho7s the ele'ents that should be in place in a safety progra'. The ele'ents are based on research and eperience fro' successful safety progra's in di:erent co'panies. &igure =. "le'ents in a safety progra' in the 23;2('odel. The Syste'ac ;ause #nalysis TechniGue is a tool to aid an invesgaon and evaluaon of incidents through the applicaon of a S;#T chart. The chart acts as a chec5list or reference to ensure that an invesgaon has loo5ed at all facets of an incident. There are Hve ersonal factors ob factors . 2nadeGuate capability ( hysicalCphysiological ( MentalCpsychological =. 3ac5 of 5no7ledge /. 3ac5 of s5ill *. Stress ( hysicalCphysiological ( MentalCpsychologica -. 2'proper 'ovaon . 2nadeGuate leadership andCor supervision =. 2nadeGuate engineering /. 2nadeGuate purchasing *. 2nadeGuate 'aintenance -. 2nadeGuate tools, eGuip'ent, 'aterials A. 2nadeGuate 7or5 standards ). ear and tear 0. #buse or 'isuse "le'ents in a safety progra' . 3eadership and ad'inistraon =. Manage'ent training /. lanned inspecon *. Tas5 analysis and procedures -. #ccidentCincident invesgaon A. Tas5 observaons ). "'ergency preparedness 0. Organisaonal rules +. #ccidentCincident analysis 1. "'ployee training . ersonal protecve eGuip'ent =. 4ealth control /. rogra' evaluaon syste' *. "ngineering controls -. ersonal co''unicaons A. Broup 'eengs ). Beneral pro'oon 0. 4iring and place'ent +. urchasing controls =1. O:(the(9ob safety Methods for accident invesgaon *- bloc5s on a S;#T chart. "ach bloc5 corresponds to a bloc5 of the loss causaon 'odel. 4ence, the Hrst bloc5 contains space to 7rite a descripon of the incident. The second bloc5 lists the 'ost co''on categories of contact that could have led to the incident under invesgaon. The third bloc5 lists the 'ost co''on i''ediate causes, 7hile the fourth bloc5 lists co''on basic causes. &inally, the boWo' bloc5 lists acvies generally accepted as i'portant for a successful loss control progra'. The techniGue is easy to apply and is supported by a training 'anual. The S;#T see's to correspond to the S6N"RB2 tool for accident registraon used in Nor7ay. #t least, the accident causaon 'odels used in S;#T and S6N"RB2 are eGuivalent. *.=.- ST" >SeGuenal 'ed events ploJng?= The ST"('ethod 7as developed by 4endric5 and @enner >+0)?. They propose a syste'ac process for accident invesgaon based on 'ul(linear events seGuences and a process vie7 of the accident pheno'ena. ST" builds on four concepts$ . Neither the accident nor its invesgaon is a single linear chain or seGuence of events.

Rather, several acvies ta5e place at the sa'e 'e. =. The event @uilding @loc5 for'at for data is used to develop the accident descripon in a 7or5sheet. # building bloc5 describes one event, i.e. one actor perfor'ing one acon. /. "vents Fo7 logically during a process. #rro7s in the ST" 7or5sheet illustrate the Fo7. *. @oth producve and accident processes are si'ilar and can be understood using si'ilar invesgaon procedures. They both involve actors and acons, and both are capable of being repeated once they are understood. ith the process concept, a speciHc accident begins 7ith the acon that started the transfor'aon fro' the described process to an = The descripon is based on 4endric5  @enner, +0). Methods for accident invesgaon *A accident process, and ends 7ith the last connected har'ful event of that accident process. The ST"(7or5sheet provides a syste'ac 7ay to organise the building bloc5s into a co'prehensive, 'ul(linear descripon of the accident process. The ST"( 7or5sheet is si'ply a 'atri, 7ith ro7s and colu'ns. There is one ro7 in the 7or5sheet for each actor. The colu'ns are labelled di:erently, 7ith 'ar5s or nu'bers along a 'e line across the top of the 7or5sheet, as sho7n in &igure ==. The 'e scale does not need to be dra7n on a linear scale, the 'ain point of the 'e line is to 5eep events in order, i.e., ho7 they relate to each other in ter's of 'e. &igure ==. ST"(7or5sheet. #n event is one actor perfor'ing one acon. #n actor is a person or an ite' that directly inFuences the Fo7 or events constung the accident process. #ctors can be involved in t7o types of changes, adapve changes or iniang changes. They can either change reacvely to sustain dyna'ic balance or they can introduce changes to 7hich other actors 'ust adapt. #n acon is so'ething done by the actor. 2t 'ay be physical and observable, or it 'ay be 'ental if the actor is a person. #n acon is so'ething that the actor does and 'ust be stated in the acve voice. The ST" 7or5sheet provides a syste'ac 7ay to organise the building bloc5s >or events? into a co'prehensive, 'ul(linear descripon of the accident process. &igure =/ sho7s an ea'ple on a #ctor # #ctor D #ctor ; #ctor @ T1 Ti'e "tc. Methods for accident invesgaon *) ST"(diagra' of an accident 7here a stone bloc5 falls o: a truc5 and hits a car/. &igure =/. #n ea'ple on a si'ple ST"(diagra' for a car accident. The ST"(diagra' in &igure =/ also sho7s the use of arro7s to lin5 tested relaonships a'ong events in the accident chain. #n arro7 convenon is used to sho7 precedeCfollo7 and logical relaons bet7een t7o or 'ore events. hen an earlier acon is necessary for a laWer to occur, an arro7 should be dra7n fro' the preceding event to the resultant event. The thought process for idenfying the lin5s bet7een events is related to the change of state concepts underlying ST" 'ethods. &or each event in the 7or5sheet, the invesgator as5s, P#re the preceding acons sucient to iniate this acons >or event? or 7ere other acons necessaryIQ Try to visuali8e the actors and acons in a P'ental 'ovieQ in order to develop the lin5s. So'e'es it is i'portant to deter'ine 7hat happened during a gap or 'e interval for 7hich 7e cannot gather any speciHc evidence. "ach re'aining gap in the 7or5sheet represents a gap in the understanding of the accident. @ac5ST" is a techniGue by 7hich you reason your 7ay bac57ards fro' the event on the right side of the 7or5sheet gap / The ST"(diagra' is based on a descripon of the accident in a ne7spaper arcle. T1 Ti'e ;ar ;ar driver Drap Stone bloc5 Truc5 Truc5 driver loads stone on truc5 Truc5 driver drives truc5 fro' # to @ Truc5 drives fro' # to @ ;ar driver dies Truc5 driver fastens the stone bloc5 ;ar drive fro' @ to # ;ar driver tries to avoid to hit the stone The car hits the stone bloc5 Stone falls o: the truc5 Drap fails The car ]collapses] >collision da'aged? ;ar driver starts the car 3egend Truc5 driver Drap fails #ctor "vent lin5 #ctor Truc5 driver ;ar driver observes the stone ;ar driver stri5es Methods for accident invesgaon *0 to7ard the event on the le side of the gap. The @ac5ST" procedure consists of as5ing a series of Phat could have led to thatIQ Guesons and 7or5ing bac57ard through the pyra'id 7ith the ans7ers. Ma5e tentave event building bloc5s for each event that ans7ers the Gueson. hen doing a @ac5ST", it is not unco''on to idenfy 'ore than one

possible path7ay bet7een the le and right events at the gap. This tells that there 'ay be 'ore than one 7ay the accident process could progress and 'ay led to develop'ent of hypothesis in 7hich should be further ea'ined. The ST"(procedure also includes so'e rigorous technical truthtesng procedures, the ro7 test, the colu'n test, and the necessaryand(sucient test. The ro7 >or hori8ontal? test tells you if you need 'ore building bloc5s for any individual actor listed along the le side of the 7or5sheet. 2t also tells you if you have bro5en each actor do7n suciently. The colu'n >or vercal? test chec5s the seGuence of events by pairing the ne7 event 7ith the acons of other actors. To pass the colu'n test, the event building bloc5 being tested 'ust have occurred  #er all the event in all the colu'ns to the le of that event,  @efore all the events in all colu'ns to the right of that event, and  #t the sa'e 'e as all the events in the sa'e colu'n. The ro7 test and the colu'n test are illustrated in &igure =*. &igure =*. or5sheet ro7 test and colu'n test. #ctor # #ctor D #ctor ; #ctor @ T1 Ti'e "tc. ;olu'ns Ro7s Ro7 tests 2s ro7 co'pleteI ;olu'n tests 2s event seGuenced Ocollision da'aged? ;ar driver starts the car Truc5 driver ;ar driver observes the stone ;ar driver bra5es @u'by road due to lac5 of 'aintenance 2naWenve car driver A ) 0 Narro7 road oor bra5es + 1 3ac5 of airbag Methods for accident invesgaon - invesgaon. The 'ethod is based on 4"S >4u'an erfor'ance "nhance'ent Syste'? 7hich is 'enoned in Table =, but not described further in this report. The MTO( analysis is based on three 'ethods$ . Structured analysis by use of an event( and cause(diagra'). =. ;hange analysis by describing ho7 events have deviated fro' earlier events or co''on pracce0. /. @arrier analysis by idenfying technological and ad'inistrave barriers in 7hich have failed or are 'issing+. &igure =A illustrates the MTO(analysis 7or5sheet. The Hrst step in an MTO(analysis is to develop the event seGuence longitudinally and illustrate the event seGuence in a bloc5 diagra'. 2denfy possible technical and hu'an causes of each event and dra7 these vercally to each event in the diagra'. &urther, analyse 7hich technical, hu'an or organisaonal barriers that have failed or 7as 'issing during the accident progress. 2llustrate all 'issing or failed barriers belo7 the events in the diagra'. #ssess 7hich deviaons or changes in 7hich di:er the accident progress fro' the nor'al situaon. These changes are also illustrated in the diagra' >see &igure =A?. The basic Guesons in the analysis are$  hat 'ay have prevented the connuaon of the accident seGuenceI  hat 'ay the organisaon have done in the past in order to prevent the accidentI The last i'portant step in the MTO( analysis is to idenfy and present reco''endaons. The reco''endaons should be as realisc and speciHc as possible, and 'ight be technical, hu'an or organisaonal. ) See subsecon *... 0 See subsecon *../. + See subsecon *..=. Methods for accident invesgaon -= &igure =A. MTO(analysis 7or5sheet. # chec5list for idenHcaon of failure causes >Pfelorsa5erQ? is also part of the MTO( 'ethodology >@ento, +++?. The chec5list contains the follo7ing factors$ . Organisaon =. or5 organisaon /. or5 pracce *. Manage'ent of 7or5 -. ;hange procedures A. "rgono'ic C deHciencies in the technology ). ;o''unicaon 0. 2nstruconsCprocedures +. "ducaonCco'petence 1. or5 environ'ent ;hange analysis "vents and causes chart @arrier analysis  Nor'al Deviaon =  >;hain of events? >;auses? Methods for accident invesgaon -/ &or each of these failure causes, there is a detailed chec5list for basic or funda'ental causes >Pgrundorsa5erQ?. "a'ples on basic causes for the failure cause 7or5 pracce are$  Deviaon fro' 7or5 instrucon  oor preparaon or planning  3ac5 of self inspecon  Use of 7rong eGuip'ent  rong use of eGuip'ent *.=.) #ccident #nalysis and @arrier &uncon >#"@? Method=1 The #ccident "voluon and @arrier &uncon >#"@? 'odel provides a 'ethod for analysis of incidents and accidents that 'odels the evoluon to7ards an incidentCaccident as a series of interacons bet7een hu'an and technical syste's. The interacon consists of failures, 'alfuncons or errors that could lead to or have resulted in an accident. The 'ethod forces analysts to integrate hu'an and technical syste's si'ultaneously 7hen perfor'ing an accident analysis starng 7ith the si'ple Fo7 chart techniGue of the 'ethod. The Fo7 chart inially consists of e'pty boes in t7o parallel colu'ns, one for the hu'an syste's and one for the technical syste's. &igure =) provides an illustraon of this diagra'. During the analysis these error boes are idenHed as the failures,

'alfuncons or errors that constute the accident evoluon. 2n general, the seGuence of error boes in the diagra' follo7s the 'e order of events. @et7een each pair of successive error boes there is a possibility to arrest the evoluon to7ards an incidentCaccident. @arrier funcon syste's >e.g. co'puter progra's? that are acvated can arrest the evoluon through e:ecve barrier funcons >e.g. the co'puter 'a5ing an incorrect hu'an intervenon 'odelled in the net error bo i'possible through bloc5ing a control?. &actors that have an inFuence on hu'an perfor'ance have been called perfor'ance shaping factors >by S7ain and BuW'an, +0/?. "a'ples of such factors are alcohol, lac5 of sleep and stress. 2n applicaon of the #"@ 'odel those factors are included in the Fo7 =1 The descripon is based on Svensson, =111. Methods for accident invesgaon -* diagra' only as S&s and they are analysed aer the diagra' has been co'pleted. S&s are included in the Fo7 diagra' in cases 7here it is possible that the factor could have contributed to one or 'ore hu'an error events. &actors such as alcohol and age are 'odelled as S&s, but never as hu'an error events or failing barrier funcons. Organisaonal factors 'ay be integrated as a barrier funcon 7ith failing or inadeGuate barrier funcons. Organisaonal factors should al7ays be treated in a special 7ay in an #"@ analysis because they include both hu'an and technical syste's. &igure =). 2llustraon of an #"@ analysis. #n #"@ analysis consists of t7o 'ain phases. The Hrst phase is to 'odel the accident evoluon in a Fo7 diagra'. 2t is i'portant to re'e'ber that #"@ only 'odels errors and that it is not an event seGuence 'ethod. #rro7s lin5 the error event boes together in order to sho7 the evoluon. The course of events is described in an approi'ate chronological order. 2t is not allo7ed to let 'ore than one arro7 lead to an error bo or to have 'ore than one arro7 going fro' a bo. The second phase consists of the barrier funcon analysis. 2n this phase, the barrier funcons are idenHed >ine:ecve andCor non eistent?. # barrier funcon represents a funcon that can arrest the accident evoluon so that the net event in the chain 7ill not be realised. # barrier funcon is al7ays idenHed in relaon to the syste's it protects, protected or could have protected. @arrier funcon syste's are the syste's perfor'ing the barrier funcons. @arrier funcon syste's can be an operator, an instrucon, a physical 4u'an error event  Technical error event  4u'an error event / 4u'an error event = #ccident C incident Technical error event = 4u'an factors syste' Technical syste' ;o''ents S& erfor'ance shaping factors &ailing orCand possible barrier funcon 3egend "rror event bo #ccidentCincident #rro7s describing the accident evoluon ossible barrier funcons ":ecve barrier funcon S& erfor'ing shaping factors Methods for accident invesgaon -- separaon, an e'ergency control syste', other safety(related syste's, etc. The sa'e barrier funcon can be perfor'ed by di:erent barrier funcon syste's. ;orrespondingly, a barrier funcon syste' 'ay perfor' di:erent barrier funcons. #n i'portant purpose of the #"@(analysis is to idenfy bro5en barrier funcons, the reasons for 7hy there 7ere no barrier funcons or 7hy the eisng ones failed, and to suggest i'prove'ents. @arrier funcons belong to one of the three 'ain categories$  2ne:ecve barrier funcons \ barrier funcons that 7ere ine:ecve in the sense that they did not prevent the develop'ent to7ard an accident  Non(eisng barrier funcons \ barrier funcons that, if present, 7ould have stopped the accident evoluon.  ":ecve barrier funcons \ barrier funcons that actually prevented the progress to7ard an accident. 2f a parcular accident should happen, it is necessary that all barrier funcons in the seGuence are bro5en and ine:ecve. The ob9ecve of an #"@( analysis is to understand 7hy a nu'ber of barrier funcons failed, and ho7 they could be reinforced or supported by other barrier funcons. &ro' this perspecve, idenHcaon of a root(cause of an accident is 'eaningless. The starng point of the analysis cannot be regarded as the root cause because the re'oval of any of all the other errors in the accident evoluon 7ould also eli'inate the accident. 2t is so'e'es dicult to 5no7 if an error should be 'odelled as an error or as a failing barrier funcon. #s

a rule of thu'b, 7hen uncertain the analysts should choose a bo and not a barrier funcon representaon in the inial #"@(analysis. The barrier funcon analysis phase 'ay be used for 'odelling of subsyste's interacons that cannot be represented seGuenally in #"@. #ll barriers funcon failures, incidents and accidents ta5e place in 'an \ technology \ organisaons contets. Therefore, an #"@( analysis also includes issues about the contet in 7hich the accident too5 place. Therefore, the follo7ing Guesons have to be ans7ered$ Methods for accident invesgaon -A . To increase safety, ho7 is it possible to change the organisaon, in 7hich the failure or accident too5 placeI =. To increase safety, ho7 is it possible to change the technical syste's contet, in 7hich the failure or accident too5 placeI 2t is i'portant to bear in 'ind that 7hen changes are 'ade in the organisaonal and technical syste's at the contet level far reaching e:ects 'ay be aWained. *.=.0 TR2OD= The 7hole research into the TR2OD concept started in +00 7hen a study that 7as contained in the report PTR2OD, # principled basis for accident prevenonQ >Reason et al, +00? 7as presented to Shell 2nternaonale etroleu' Maatschappi9, "ploraon and roducon. The idea behind TR2OD is that organisaonal failures are the 'ain factors in accident causaon. These factors are 'ore PlatentQ and, 7hen contribung to an accident, are al7ays follo7ed by a nu'ber of technical and hu'an errors. The co'plete TR2OD( 'odel== is illustrated in &igure =0. &igure =0. The co'plete TR2OD 'odel. Substandard acts and situaons do not 9ust occur. They are generated by 'echanis's acng in organisaons, regardless 7hether there has been an accident or not. Oen these 'echanis's result fro' decisions = This descripon is based on Broene7eg, ++0. == The TR2OD('odel described here 'ight be di:erent fro' previously published 'odels based on the TR2OD theory, but this 'odel is fully co'pable 7ith the 'ost resent version of the accident invesgaon tool TR2OD @eta described later in this chapter. Decision 'a5ers Substandard acts sychological precursors Operaonal disturbance ;onseGuences #ccident @reached barriers @reached barriers 3atent failures 1 @R&s 3atent failures @R& Defences Methods for accident invesgaon -) ta5en at high level in the organisaon. These underlying 'echanis's are called @asic Ris5 &actors=/ >@S&s?. These @S&s 'ay generate various psychological precursors in 7hich 'ay lead to substandard acts and situaons. "a'ples on psychological precursors of slips, lapses and violaons are 'e pressure, being poorly 'ovated or depressed. #ccording to this 'odel, eli'inang the latent failures categori8ed in @R&s or reducing their i'pact 7ill prevent psychological precursors, substandard acts and the operaonal disturbances. &urther'ore, this 7ill result in prevenon of accidents. The idenHed @R&s cover hu'an, organisaonal and technical proble's. The di:erent @asic Ris5 &actors are deHned in Table -. Ten of these @R&s leading to the Poperaonal disturbanceQ >the PprevenveQ @R&s?, and one @R& is ai'ed at controlling the conseGuences once the operaonal disturbance has occurred >the P'igaonQ @R&?. There are Hve generic prevenon @R&s >A \ 1 in Table -? and Hve speciHc @R&s > \ - in Table -?. The speciHc @R&s relate to latent failures that are speciHc for the operaons to be invesgated >e.g. the reGuire'ents for Tools and "Guip'ent are Guite di:erent in a oil drilling environ'ent co'pared to an intensive care 7ard in a hospital?. These  @R&s have been idenHed as a result of brainstor'ing, a study of audit reports, accident scenarios, a theorecal study, and a study on o:shore pla^or's. The division is deHnive and has sho7n to be valid for all industrial applicaons. =/ These 'echanis's 7ere inially called Beneral &ailure Types >B&Ts?. Methods for accident invesgaon -0 Table -. The deHnions of the basic ris5 factors >@R&s? in TR2OD. No @asic Ris5 &actor #bbr. DeHnion  Design D" "rgono'ically poor design of tools or eGuip'ent >user( unfriendly? = Tools and eGuip'ent T" oor Guality, condion, suitability or availability of 'aterials, tools, eGuip'ent and co'ponents / Maintenance 'anage'ent MM No or inadeGuate perfor'ance of 'aintenance tas5s and repairs * 4ouse5eeping 4< No or insucient aWenon given to 5eeping the 7or5

Foor clean or died up - "rror enforcing condions "; Unsuitable physical perfor'ance of 'aintenance tas5s and repairs A rocedures R 2nsucient Guality or availability of procedures, guidelines, instrucons and 'anuals >speciHcaons, Ppaper7or5Q, use in pracce? ) Training TR No or insucient co'petence or eperience a'ong e'ployees >not suciently suitedCinadeGuately trained? 0 ;o''unicaon ;O No or ine:ecve co''unicaon bet7een the various sites, depart'ents or e'ployees of a co'pany or 7ith the ocial bodies + 2nco'pable goals 2B The situaon in 7hich e'ployees 'ust choose bet7een op'al 7or5ing 'ethods according to the established rules on one hand, and the pursuit of producon, Hnancial, polical, social or individual goals on the other 1 Organisaon OR Shortco'ings in the organisaonEs structure, organisaonEs philosophy, organisaonal processes or 'anage'ent strategies, resulng in inadeGuate or ine:ecve 'anage'ent of the co'pany  Defences D& No or insucient protecon of people, 'aterial and environ'ent against the conseGuences of the operaonal disturbances TR2OD @eta The TR2OD @eta(tool is a co'puter(based instru'ent that provides the user 7ith a tree(li5e overvie7 of the accident that 7as invesgated. 2t is a 'enu driven tool that 7ill guide the invesgator through the process of 'a5ing an electronic representaon of the accident. Methods for accident invesgaon -+ The @"T#(tool 'erges t7o di:erent 'odels, the 4"M >The 4a8ard and ":ects Manage'ent rocess? 'odel and the TR2OD 'odel. The 'erge has resulted in an incident causaon 'odel that di:ers conceptually fro' the original TR2OD 'odel. The 4"M 'odel is presented in &igure =+. &igure =+. P#ccident 'echanis'Q according to 4"M. The TR2OD @eta accident causaon 'odel is presented in &igure /1. This string is used to idenfy the causes that lead to the breaching of the controls and defences presented in the 4"M 'odel. &igure /1. TR2OD @eta #ccident ;ausaon Model. #lthough the 'odel presented in &igure /1 loo5s li5e the original TR2OD 'odel, its co'ponents and assu'pons are di:erent. 2n the @eta('odel the defences and controls are directly lin5ed to unsafe acts, precondions and latent failures. Unsafe acts describe ho7 the barriers 7ere breached and the latent failures 7hy the barriers 7ere breached. #n ea'ple of a TR2OD @eta accident analysis is sho7n in &igure /. 4a8ard #ccidentC event %ic' or target &ailed control &ailed defence #ccident &ailed controls or defences 3atent failure>s? recondion>s? #cve failure>s? Methods for accident invesgaon A1 &igure /. "a'ple on a TR2OD @eta analysis. The ne7 7ay of invesgang accidents >see &igure /=? is Guite di:erent fro' the convenonal ones. No research is done to idenfy all the contribung substandard acts or clusters of substandard acts, the target for invesgaon is to Hnd out 7hether any of the @asic Ris5 &actors are acng. hen the @R&s have been idenHed, their i'pact can be decreased or even be eli'inated. The real source of proble's is tac5led instead of the sy'pto's. 4a8ard$ ointed table corner "'ployee hits table. 2n9ured 5nee %ic' oor ha8ard register oor ha8ard register Missing control Rounded or rubber corners Missing control #udit for obstacles Missing defence the targets of control?, and the up7ard Fo7 of state infor'aon >the 'easure'ents of control?. Decision recondion Order &uncon lan Decision Order 2ndirect conseGuence Tas5 or #con Tas5 or acon Direct conseGuence ;onseGuence recondion evaluated no further riories Syste' level . Bovern'ent. olicy  budgeng =. Regulatory bodies and #ssociaons /. 3ocal area govern'ent ;o'pany 'anage'ent lanning  budgeng ;rical event 3oss of control or loss of contain'ent Direct conseGuence *. Technical  operaonal 'anage'ent -. hysical processes  #ctor acvites A. "Guip'ent  surroundings 0  -  Reference to annotaons 2nFuence Methods for accident invesgaon A- &igure /-. rincipal illustraon of an #ctorMap. #ctor Syste' level . Bovern'ent =. Regulatory bodies /. Regional  3ocal govern'ent ;o'pany 'anage'ent *. Technical  operaonal 'anage'ent #ssociaons -. Operators #ctor #ctor #ctor #ctor #ctor #ctors #ctor #ctor #ctor #ctor #ctor #ctor #ctor #ctor #ctor #ctor Methods for accident invesgaon A) - Discussion and conclusion -. Discussion ithin the Held of accident invesgaon, there are no co''on agree'ent of deHnions of concepts, it tend to be a liWle confusion of ideas. "specially the noon of cause has been discussed. hile so'e invesgators focus on causal factors >e.g. DO", ++)?, others focus on deter'ining factors >e.g. e.g. 4op5ins, =111?, acve failures and latent condions >e.g. Reason, ++)? or safety proble's >4endric5  @enner, +0)?. @R&? >Pevent triosQ?. The third colu'n covers the level of scope of the di:erent analysis 'ethods. The levels correspond to the di:erent levels of the sociotechnical syste' involved in ris5 'anage'ent illustrated in &igure *. The di:erent levels are$ . The 7or5 and technological syste' =. The sta: level /. The 'anage'ent level *. The co'pany level -. The regulators and associaons level A. The Bovern'ent level #s sho7n in Table A, the scope of 'ost of the 'ethods is li'ited to level  \ *. #lthough ST" 7as originally developed to cover level  \ *, eperience fro' S2NT"&Es accident invesgaons sho7s that the Methods for accident invesgaon A+ 'ethod also 'ay be used to analyse events inFuenced by the regulators and the Bovern'ent. 2n addion to ST", only #cci(Map put focus on level - and A. This 'eans that invesgators focusing on the Bovern'ent and the regulators in their accident invesgaon to a great etend need to base their analysis on eperience and praccal 9udge'ent 'ore than on results fro' for'al analysis 'ethods. The fourth colu'n states 7hether the 'ethods are a pri'ary 'ethod or a secondary 'ethod. ri'ary 'ethods are stand(alone techniGues 7hile secondary 'ethods provide special input as supple'ent to other 'ethods. "vents and causal factors charng, ST", MTO( analysis, TR2OD and #cci('ap are all pri'ary 'ethods. The fault tree analysis and event tree analysis 'ight be both pri'ary and secondary 'ethods. The other 'ethods are secondary 'ethods. 2n the Hh colu'n the di:erent 'ethods are categori8ed as deducve, inducve, 'orphological or non(syste' oriented. &ault tree analysis and MORT are deducve 'ethods 7hile event three analysis is an inducve 'ethod. #cci('ap 'ight be both inducve and deducve. The #"@('ethod is characteri8ed as 'orphological, 7hile the other 'ethods are non(syste' oriented. 2n the sith colu'n the 'ethods are lin5ed to di:erent types of accident 'odels in 7hich have inFuenced the 'ethods. The follo7ing accident 'odels are used$ # ;ausal(seGuence 'odel @ rocess 'odel ; "nergy 'odel D 3ogical tree 'odel " S4"('anage'ent 'odels Root cause analysis, S;#T and TR2OD are based on causal(seGuence 'odels. "vents and causal charng, change analysis, events and causal factors analysis, ST", MTO( analysis and #"@('ethod are based on process 'odels. The barrier analysis is based on the energy 'odel. &ault tree analysis, event tree analysis and MORT are based on logical tree 'odels. MORT and S;#T are also based on S4"('anage'ent 'odels. The #cci('ap is based on a co'binaon of a causal( seGuence 'odel, a process 'odel and a logical tree 'odel. Methods for accident invesgaon )1 2n the last colu'n, there is 'ade an assess'ent of the need of educaon and training in order to use the 'ethods. The ter's P"pertQ, PSpecialistQ and PNoviceQ are used in the table. "pert indicates that there is need of for'al educaon and training before people are able to use the 'ethods in a proper 7ay. So'e eperience is also beneHcial. &ault tree analysis, MORT and #cci('ap enter into this category. Novice indicates that people are able to use the 'ethods aer and orientaon of the 'ethods 7ithout

hands(on training or eperience. "vents and causal factors charng, barrier analysis, change analysis and ST" enter into this category. Specialist is so'e7here bet7een epert and novice and events and causal factors analysis, root cause analysis, event tree analysis, S;#T, MTO(analysis, #"@'ethod and TR2OD enter into this category. Methods for accident invesgaon ) Table A. ;haracteriscs of di:erent accident invesgaon 'ethods. Method #ccident seGuence 3evels of analysis ri'ary C secondary #nalycal approach #ccident 'odel Training need "vents and causal factors charng 6es (* ri'ary Non(syste' oriented @ Novice @arrier analysis No (= Secondary Non(syste' oriented ; Novice ;hange analysis No (* Secondary Non(syste' oriented @ Novice "vents and causal factors analysis (* Secondary Non(syste' oriented @ Specialist Root cause analysis No (* Secondary Non(syste' oriented # Specialist &ault tree analysis No (= ri'aryC Secondary Deducve D "pert "vent Tree analysis No (/ ri'aryC Secondary 2nducve D Specialist MORT No =(* Secondary Deducve D C " "pert S;#T No (* Secondary Non(syste' oriented # C " Specialist ST" 6es (A ri'ary Non(syste' oriented @ Novice MTOanalysis 6es (* ri'ary Non(syste' oriented @ SpecialistC epert #"@'ethod No (/ Secondary Morphological @ Specialist TR2OD 6es (* ri'ary Non(syste' oriented # Specialist #cci(Map No (A ri'ary Deducve  inducve # C @ C D "pert -.= ;onclusion Ma9or accidents al'ost never result fro' one single cause, 'ost accidents involve 'ulple, interrelated causal factors. #ll actors or decision( 'a5ers inFuencing the nor'al 7or5 process 'ight also Methods for accident invesgaon )= inFuence accident scenarios, either directly or indirectly. This co'pleity should also reFect the accident invesgaon process. The ai' of accident invesgaons should be to idenfy the event seGuences and all >causal? factors inFuencing the accident scenario in order to be able to suggest ris5 reducing 'easures in 7hich 'ay prevent future accidents. This 'eans that all 5ind of actors, fro' technical syste's and front(line operators to regulators and the Bovern'ent need to be analysed. Oen, accident invesgaons involve using of a set of accident invesgaon 'ethods. "ach 'ethod 'ight have di:erent purposes and 'ay be a liWle part of the total invesgaon process. Re'e'ber, every piece of a pu88le is as i'portant as the others. Braphical illustraons of the event seGuence are useful during the invesgaon process because it provides an e:ecve visual aid that su''aries 5ey infor'aon and provide a structured 'ethod for collecng, organising and integrang collected evidence to facilitate co''unicaon a'ong the invesgators. Braphical illustraons also help idenfying infor'aon gaps. During the invesgaon process di:erent 'ethods should be used in order to analyse arising proble' areas. #'ong the 'ul(disciplinary invesgaon tea', there should be at least one 'e'ber having good 5no7ledge about the di:erent accident invesgaon 'ethods, being able to choose the proper 'ethods for analysing the di:erent proble's. ust li5e the 'echanicians have to choose the right tool on order to repair a technical syste', an accident invesgator has to choose proper 'ethods analysing di:erent proble' areas. Methods for accident invesgaon )/ A References #ndersson R.  Menc5el "., ++-. On the prevenon of accidents and in9uries. # co'parave analysis of conceptual fra'e7or5s. #ccident #nalysis and revenon, %ol. =), No. A, )-) \ )A0. #rbeids'il9Ysenteret, =11. %eiledning i uly55esgrans5ing, #rbeids'il9Yforlaget, =11 @ento, (., +++. MTO(analys av hndelsesrapporter, OD( 11(= @ird, &.". r  Ber'ain, B.3., +0-. raccal 3oss ;ontrol 3eadership. 2S@N 1(001A(1-*(+, 2nternaonal 3oss ;ontrol 2nstute, Beorgia, US#. ;;S, ++=. Buidelines for 2nvesgang ;he'ical rocess 2ncidents. 2S@N 1(0A+(1---(, ;enter for ;he'ical rocess Safety of the #'erican 2nstute of ;he'ical "ngineers, ++=. DO", ++). 2'ple'entaon Buide &or Use ith DO" Order ==-.#, #ccident 2nvesgaons, DO" B ==-.#( Nove'ber =A, ++)CRev. , U.S. Depart'ent of "nergy, ashington D.;, US#. DO", +++. ;onducng #ccident 2nvesgaons DO" or5boo5, Revision =, May , +++, U.S. Depart'ent of "nergy, ashington D.;, US#. &erry T.S., +00. Modern accident invesgaon and

analysis >=nd ed.?. 2S@N 1.*)(A=*0(1, iley 2nterscience publicaon, United States. Bil9e, N. og Bri'en, 4., ++/. Sa'funnsvitens5apenes forutsetninger 2nnfYring i sa'funnsvitens5apenes vitens5apsHlosoH, Universitetsforlaget, Oslo. Broene7eg, ., ++0. ;ontrolling the controllable The 'anage'ent of safety. &ourth edion. DSO ress, 3eiden University, The Netherlands, ++0. 4ale #, ilpert @, &reitag M, ++). #er the event ( fro' accident to organisaonal learning. 2S@N 1 10 1*/1)*1, erga'on, ++). 4endric5 #"@? Method ......... -/ *.=.0 TR2OD................................................................................... -A *.=.+ #cci( 'ap.................................................................................. A * - D2S;USS2ON #ND ;ON;3US2ON................................................ A) -. D2S;USS2ON.................................................................................. A) -.= ;ON;3US2ON................................................................................ ) A R"&"R"N;"S ................................................................................... )/ Methods for accident invesgaon -  2ntroducon . 2ntroducon to accident invesgaon and deli'itaons of the report The accident invesgaon process consists of a 7ide range of acvies, and is described so'e7hat di:erent by di:erent authors. DO" >+++? divide the invesgaon process into three phasesK collecon of evidence and facts, analysis of these facts, and develop'ent of conclusions and develop'ent of 9udg'ents of need and 7ring the report, see &igure . These are all overlapping phases and the 7hole process is iterave. So'e authors also include the i'ple'entaon and follo7(up of reco''endaons in the invesgaon phase >e.g., see &igure =? $  This approach is not li'ited to 'a9or accidents, but also include occupaonal accidents. ;ollecon of evidence and facts #nalysis of evidence and factsK Develop'ent of conclusions Develop'ent of 9udg'ents of needK ring the report Methods for accident invesgaon A . #ll reported incidents >accidents and near accidents? are invesgated i''ediately at the Hrst level by the supervisor and safety representave. =. # selecon of serious incidents, i.e. freGuently recurring types of incidents and incidents 7ith high loss potenal >actual or possible? are subseGuently invesgated by a proble'(solving group. /. On rare occasions, 7hen the actual or potenal loss is high, an accident invesgaon co''ission carries out the invesgaon. This co''ission has an independent status in relaon to the organisaons that are responsible for the occurrence. &igure =. #ccident invesgaon at three levels >Reason, ++)?. Organisaonal accidents are the co'paravely rare, but oen catastrophic, events that occur 7ithin co'ple, 'odern technologies such as nuclear po7er plants, co''ercial aviaon, petroche'ical industry, etc. Organisaonal accidents have 'ulple causes involving 'any people operang at di:erent levels of their respecve co'panies. @y contrast, individual accidents are accidents in 7hich a speciHc person or a group is oen both the agent and the vic' of the accident. Organisaonal accidents 2ndependant invesgaon co''ission or5 place roble'(solving group 2''ediate invesgaon by Hrst(line supervisor Reporng 2'ple'entaon of re'edial acons #ccidents Near accidents #ll events #ll events 2n eceponal cases &reGuent or severe events Methods for accident invesgaon ) are according to Reason >++)? a product of technological innovaons that have radically altered the relaonship bet7een syste's and their hu'an ele'ents. Ras'ussen >++)? proposes di:erent ris5 'anage'ent strategies for di:erent 5inds of

accidents, see &igure /. The accident invesgaon 'ethods dealt 7ith in this report are li'ited to 'ethods used for evoluonary safety control, i.e. in(depth analysis of 'a9or accidents >ref. e.g., DO", ++)?, others focus on deter'ining factors >e.g., e.g., 4op5ins, =111?, acve failures and latent condions >e.g., Reason, ++)? or safety proble's >4endric5  @enner, +0)?. 4op5ins >=111? deHnes cause in the follo7ing 7ay$ Pone thing is said to be a cause of another if 7e can say but for the Hrst the second 7ould not have occurredQ. 3eplat >++)? epresses this in a 'ore for'al 7ay by saying that in general, the follo7ing type of deHnion of cause is accepted$ Pto say that event  is the cause of event 6 is to say that the accidentQ period instead of on the 'igaon of the conseGuence of the accident. Methods for accident invesgaon + occurrence of  is a necessary condion to the producon of 6, in the circu'stances consideredQ. Such a deHnion i'plies that if any one of the causal path7ays idenHed are re'oved, the outco'e 7ould probably not have occurred. Using the ter' contribung factor 'ay be less for'al, if an event has not occurred, this 7ould necessarily not prevented the occurrence of the accident. =11? reco''ends avoiding the 7ord cause in accident invesgaons and rather tal5 about 7hat 'ight have prevented the accident. #ccident invesgators 'ay use di:erent fra'es for their analysis of accidents, but nevertheless the conclusions about 7hat happened, 7hy did it happen and 7hat 'ay be done in order to prevent future accidents 'ay be the sa'e. So'e deHnions are included in this chapter. These deHnions are 'eant as an introducon to the ter's. Several of the ter's are deHned in di:erent 7ays by di:erent authors. The deHnions are Guoted 7ithout any co''ents or discussions in

this report in order to sho7 so'e of the specter. Therefore, these deHnions represent the authorsE opinions. #ccident # seGuence of logically and chronologically related deviang events involving an incident that results in in9ury to personnel or da'age to the environ'ent or 'aterial assets. >DO", ++)? @arrier #nything used to control, prevent, or i'pede energy Fo7s. ;o''on types of barriers include eGuip'ent, ad'inistrave procedures and processes, supervisionC'anage'ent, 7arning devices, 5no7ledge and s5ills, and physical. @arriers 'ay be either control or safety. >DO", ++)? @arrier analysis #n analycal techniGue used to idenfy the energy sources and the failed or deHcient barriers and controls that contributed to an accident. >DO", ++)? Methods for accident invesgaon 1 ;ausal factor #n event or condion in the accident seGuence necessary and sucient to produce or contribute to the un7anted result. ;ausal factors fall into three categoriesK direct cause, contribung cause and root cause. >DO", ++)? ;ause of accident ;ontribung factor or root cause. >DO", ++)? ;ontribung factor More lasng ris5(increasing condion at the 7or5place related to design, organisaon or social syste'. >;;S, ++=? The causal factor>s? that, if corrected, 7ould prevent recurrence of the accident. >DO", ++)? Most basic cause of an accidentCincident, i.e. a lac5 of adeGuate 'anage'ent control resulng in deviaons and contribung factors. >bla'e?  "valuate the Gueson of guilt in order to assess the liability for co'pensaon >pay? #s 7e see, there 'ay be di:erent purposes in 7hich iniate accident invesgaons. The di:erent purposes 7ill not be discussed any'ore in this report. =.* Responsibility for accident invesgaon ho should be responsible for perfor'ing accident invesgaons and ho7 thoroughly should the accident be invesgatedI The history of accident invesgaon in the past decades sho7s a trend to go further and further bac5 in the analysis, i.e., fro' being sasHed 7ith idenfying hu'an errors by front(personnel or technical failures to idenfy 7ea5nesses in the govern'ental policies as root causes. 2n order to 5no7 7hen 7e should stop our invesgaon, 7e need 7hat Ras'ussen >++1? called stop(rules. Reason >++)? suggests that 7e should stop 7hen the causes idenHed are no longer controllable. The stopping rule suggested by Reason >++)?, leads to di:erent stopping points for di:erent pares. ;o'panies should trace causes bac5 to failures in their o7n 'anage'ent syste's and develop ris5reducing 'easures that they have authority to i'ple'ent. Supervisory authories >e.g., The Nor7egian etroleu' Directorate?, appointed govern'ental co''issions of inGuiries >e.g., the Sleipnerco''ission, and the Vsta(co''ission? or per'anent invesgaon boards >e.g., The Nor7egian #ircra #ccident 2nvesgaon @oard? Methods for accident invesgaon A should in addion focus on regulatory syste's and as5 7hether 7ea5nesses in these syste's contributed to the accident. The police and the prosecung authority are responsible for evaluang the basis for potenal cri'inal prosecuon, 7hile the court of 9usce is responsible for passing sentence on a person or a co'pany. The liability for co'pensaon is 7ithin the insurance co'paniesE and the la7yerEs range of responsibility. =.- ;riteria for accident invesgaons hat is a PgoodQ accident invesgaonI This Gueson is dicult to ans7er in a si'ple 7ay, because the ans7er depends on the purpose of the invesgaon. Nevertheless, 2 have included ten funda'ental criteria for

accident invesgaons stated by 4endric5  @enner >+0)?. Three criteria are related to ob9ecves and purposes of the accident invesgaon, four to invesgave procedures, and three to the outputs fro' the invesgaon and its usefulness. ;riteria related to ob9ecves and purposes  Realisc The invesgaon should result in a realisc descripon of the events that have actually occurred.  Non( causal #n invesgaon should be conducted in a non(causal fra'e7or5 and result in an ob9ecve descripon of the accident process events. #Wribuon of cause or fault can only be considered separate fro', and aer the understanding of the accident process is co'pleted to sasfy this criterion.  ;onsistent The invesgaon perfor'ance fro' accident to accident and a'ong invesgaons of a single accident to di:erent invesgators should be consistent. Only consistency bet7een results of di:erent invesgaons enables co'parison bet7een the'. Methods for accident invesgaon ) ;riteria related to invesgaon procedures  Disciplining #n invesgaon process should provide an orderly, syste'ac fra'e7or5 and set of procedures to discipline the invesgatorsE tas5s in order to focus their e:orts on i'portant and necessary tas5s and avoid duplicave or irrelevant tas5s.  &unconal #n invesgaon process should be funconal in order to 'a5e the 9ob ecient, e.g. by helping the invesgator to deter'ine 7hich events 7ere part of the accident process as 7ell as those events that 7ere unrelated.  DeHnive #n invesgaon process should provide criteria to idenfy and deHne the data that is needed to describe 7hat happened.  ;o'prehensive #n invesgaon process should be co'prehensive so there is no confusion about 7hat happened, no unsuspected gaps or holes in the eplanaon, and no conFict of understanding a'ong those 7ho read the report. ;riteria related to output and usefulness  Direct The invesgaon process should provide results that do not reGuire collecon of 'ore data before the needed controls can be idenHed and changes 'ade.  Understandable The output should be readily understandable.  Sasfying The results should be sasfying for those 7ho inialised the invesgaon and other individuals that de'and results fro' the invesgaons. So'e of these criteria are debatable. &or instance 7ill the second criterion related to causality be disputable. 2nvesgators using the causal(seGuence accident 'odel 7ill in principle focus on causes during their invesgaon process. #lso the last criterion related to sasfacon 'ight be discussed. 2'agine an invesgaon inialised by the top 'anage'ent in a co'pany. 2f the top 'anage'ent is cricised Methods for accident invesgaon 0 in the accident report, they are not necessarily sasHed 7ith the results, but nevertheless it 'ay be a PgoodQ invesgaon. Methods for accident invesgaon + / The accident invesgaon process &igure - sho7s the detailed accident invesgaon process as described by DO" >+++?. #s sho7n in the Hgure, the process starts i''ediately 7hen an accident occurs, and the 7or5 is not Hnished before the Hnal report is accepted by the appoinng ocial. This report focuses on the process of analysing evidence to deter'ine and evaluate causal factors >see chapter *?, but Hrst a fe7 co''ents to the other 'ain phases. &igure -. DO"Es process for accident invesgaon >DO", +++?. @oard acvites #ccident occurs Develop conclusions and deter'ine 9udg'ents of need "valuate causal factors 2ntegrate, organise, and analyse evidence to deter'ine causal factors ;ollect, preserve, and verify evidence @oard chairperson ta5es responsibility for accident scene @oard arrives at accident scene #ppoinng ocial Selects @oard chairperson and 'e'bers Readiness tea' responds Secures scene Ta5es 7itness state'ents reserves evidence 2nial reporng and categorisaon ;onduct reGuire'ents veriHcaon analysis repare dra report @oard 'e'bers Hnalise dra report #ppoinng ocial accepts report Site organisaons conduct fractual accurace revie7 @oard chairperson conducts closeout brieHng Methods for accident invesgaon =1 /. ;ollecng evidence and facts ;ollecng data is a crical part of  the invesgaon. Three 5ey types of evidence are collected during the invesgaon process$  4u'an or testa'entary evidence 4u'an or testa'entary evidence includes 7itness state'ents and observaons.

 hysical evidence hysical evidence is 'aWer related to the accident >e.g. eGuip'ent, parts, debris, hard7are, and other physical ite's?.  Docu'entary evidence Docu'entary evidence includes paper and electronic infor'aon, such as records, reports, procedures, and docu'entaon. The 'a9or steps in gathering evidence are collecng hu'an, physical and docu'entary evidence, ea'ining organisaonal concerns, 'anage'ent syste's, and line 'anage'ent oversight and at last preserving and controlling the collected evidence. ;ollecng evidence can be a lengthy, 'e(consu'ing, and piece'eal process. itnesses 'ay provide s5etchy or conFicng accounts of the accident. hysical evidence 'ay be badly da'aged or co'pletely destroyed, Docu'entary evidence 'ay be 'ini'al or dicult to access. Thorough invesgaon reGuires that board 'e'bers are diligent in pursuing evidence and adeGuately eplore leads, lines of inGuiry, and potenal causal factors unl they gain a suciently co'plete understanding of the accident. This topic 7ill not be discussed any'ore in this report, but for those interested in the topic are the follo7ing references usefulK DO" >+++?, ;;S >++=? and 2ngstad >+00?. Methods for accident invesgaon = /.= #nalysis of evidence and facts #nalysis of evidence and facts is the process of deter'ining causal factors, idenfy latent condions or contribung factors >or 7hatever you 7ant to call it? and see5s to ans7er the follo7ing t7o Guesons$  hat happened 7here and 7henI  hy did it happenI DO" >+++? describes three types of causal factors$ . Direct cause =. ;ontribung causes /. Root causes # direct cause is an i''ediate event or condion that caused the accident >DO", ++)?. # contribung cause is an event or condion that together 7ith other causes increase the li5elihood of an accident but 7hich individually did not cause the accident >DO", ++)?. # root cause is the causal factor>s? that, if corrected, 7ould prevent recurrence of the accident >DO", ++)?. There are di:erent opinions of the concept of causality of accidents, see co''ents in secon .=., but this topic 7ill not be discussed any further here. ;;S >++=? lists three analycal approaches by 7hich conclusions can be reached about an accident$  Deducve approach  2nducve approach.  Morphological approach 2n addion, there eists di:erent concepts for accident invesgaon not as co'prehensive as these syste'(oriented techniGues. These are categori8ed as non(syste'(oriented techniGues. The deducve approach involves reasoning fro' the general to the speciHc. 2n the deducve analysis, it is postulated that a syste' or process has failed in a certain 7ay. Net an aWe'pt is 'ade to deter'ine 7hat 'odes of syste', co'ponent, operator and organisaon behaviour contribute to the failure. The 7hole accident Methods for accident invesgaon == invesgaon process is a typical ea'ple of a deducve reasoning. &ault tree analysis is also an ea'ple of a deducve techniGue. The inducve approach involves reasoning fro' individual cases to a general conclusion. #n inducve analysis is perfor'ed by postulang that a parcular fault or iniang event has occurred. 2t is then deter'ined 7hat the e:ects of the fault or iniang event are on the syste' operaon. ;o'pared 7ith the deducve approach, the inducve approach is an Povervie7Q 'ethod. #s such it bring an overall structure to the invesgave process. To probe the details of the causal factors, control and barrier funcon, it is oen necessary to apply deducve analysis. "a'ples of inducve techniGues are failure 'ode and e:ects analysis >&M";#?, 4#XOEs and event tree analysis. The 'orphological approach to analycal incident invesgaon is based on the structure of the syste' being studied. The 'orphological approach focuses directly on potenally ha8ardous ele'ents >for ea'ple operaon, situaons?. The ai' is to concentrate on the factors having the 'ost signiHcant inFuence on safety. hen perfor'ing a 'orphological analysis, the analyst is pri'arily applying his or her past eperience of incident invesgaon. Rather than loo5ing at all possible deviaons 7ith and 7ithout a potenal safety i'pact, the invesgaon focuses on 5no7n ha8ard sources. Typically, the 'orphological approach is an adaptaon of deducve or inducve approaches, but 7ith its o7n guidelines. S2NT"& has developed a

useful Hve(step 'odel for invesgaon of causes of accidents. The 'odel is illustrated in &igure A. Step  is idenHcaon of the event seGuences 9ust before the accident. Step = is idenHcaon of deviaons and failures inFuencing the event seGuence that led to the accident. This includes deviaons fro' eisng procedures, deviaons fro' co''on pracce, technical failures and hu'an failures. Step / is idenHcaon of 7ea5nesses and defects 7ith the 'anage'ent syste's. The ob9ecve is to detect possible causes of the deviaons or failures idenHed in Step =. Step * is idenHcaon of 7ea5nesses and defects related to the top 'anage'ent of the co'pany, because it is their responsibility to establish the necessary 'anage'ent syste's and ensure that the syste's are co'plied 7ith. Step - is idenHcaon of potenal Methods for accident invesgaon =/ deHciencies related to the public safety fra'e7or5, i.e. 'ar5ed condions, la7s and regulaons. &igure A. S2NT"&Es 'odel for analysis of accident causes >#rbeids'il9Ysenteret, =11?. Di:erent 'ethods for analysis of evidence and facts are further discussed in chapter *. DeHciencies related to the public safety fra'e7or5 Z "cono'y Z 3abour Z 3a7s and regulaons etc. "vent seGuence Z Decisions Z #cons Z O'issions Deviaons and failures inFuencing the event seGuence Z rocedures not follo7ed Z Technical failures Z 4u'an failures ea5nesses and defects 7ith the 'anage'ent syste's Z 3ac5 of or inadeGuate procedures Z 3ac5 of i'ple'entaon Z 2nsucient trainingCeducaon Z 2nsucient follo7(up ea5nesses and defects related to the top 'anage'ent Z olicy Z Organisaon and responsibilites Z 2nFuence on aJtudes Z &ollo7(up by 'anage'ent Undesirable event 3oss C in9uries on Z ersonnel Z roperes Z "nviron'ent #nalysis of causes #nalysis of conseGuences Step  Step - Step * Step / Step = #nalysis of organisaon ST"analysis Methods for accident invesgaon =* /./ Reco''endaons and reporng One of the 'ain ob9ecves of perfor'ing accidents invesgaons is to idenfy reco''endaons that 'ay prevent the occurrence of future accidents. This topic 7ill not be discussed any further, but the reco''endaons should be based on the analysis of evidence and facts in order to prevent that the revealed direct and root causes 'ight lead to future accidents. #t the co'pany level the reco''ended ris5 reducing 'easures 'ight be focused on technical, hu'an, operaonal andCor organisaonal factors. Oen, it is even 'ore i'portant to focus aWenon to7ards changes in the higher levels in &igure *, e.g., by changing the regulaons or the authoritave supervisory pracce. # useful p is to be open('inded in the search for ris5 reducing 'easures and not to be narro7 in this part of the 7or5. 4endric5 and @enner >+0)? says that t7o thoughts should be 5ept in 'ind regarding accident reports$  2nvesgaons are re'e'bered trough their reports  The best invesgaon 7ill be 7asted by a poor report. Methods for accident invesgaon =- * Methods for accident invesgaons # nu'ber of 'ethods for accident invesgaon have been developed, 7ith their o7n strengths and 7ea5nesses. So'e 'ethods of great i'portance are selected for further ea'inaon in this chapter. The selecon of 'ethods for further descripon is not based on any scienHc selecon criteria. @ut the 'ethods are 7idely used in pracce, 7ell ac5no7ledged, 7ell described in the literature* and so'e 'ethods that are relavely recently developed. 2n order to sho7 the span in di:erent accident invesgaon 'ethods, Table  sho7s an oversight over 'ethods described by DO" >+++? and Table = sho7s an oversight described by ;;S >++=?. So'e of the 'ethods in the tables are overlapping, 7hile so'e are di:erent. Table . #ccident invesgaon analycal techniGues presented in DO" >+++?. ;ore #nalycal TechniGues "vents and ;ausal &actors ;harng and #nalysis @arrier #nalysis ;hange #nalysis Root ;ause #nalysis ;o'ple #nalycal TechniGues &or co'ple accidents 7ith 'ulple syste' failures, there 'ay in addion be need of analycal techniGues li5e analyc tree analysis, e.g. &ault Tree #nalysis MORT >Manage'ent Oversight and Ris5 Tree? "T >ro9ect "valuaon Tree #nalysis? SpeciHc #nalycal TechniGues 4u'an &actors #nalysis 2ntegrated #ccident "vent Matri &ailure Modes and ":ects #nalysis So7are 4a8ards #nalysis ;o''on ;ause &ailure

#nalysis Snea5 ;ircuit #nalysis )=(4our roHle Materials and Structural #nalysis ScienHc Modelling >e.g., for incidents involving cricality and at'ospheric despersion? * So'e 'ethods are co''ercialised and therefore li'ited described in the public available literature. Methods for accident invesgaon =A Table =. #ccident invesgaons 'ethods described by ;;S >++=?. 2nvesgaon 'ethod #ccident #nato'y 'ethod >##M? #con "rror #nalysis >#"#? #ccident "voluon and @arrier #nalysis >#"@? ;hange "valuaonC#nalysis ;ause(":ect 3ogic Diagra' >;"3D? ;ausal Tree Method >;TM? &ault Tree #nalysis >&T#? 4a8ard and Operability Study >4#XO? 4u'an erfor'ance "nhance'ent Syste' >4"S? 4u'an Reliability #nalysis "vent Tree >4R#("T? Mulple(;ause, Syste's(oriented 2ncident 2nvesgaon >M;SO22? Mullinear "vents SeGuencing >M"S? Manage'ent Oversight Ris5 Tree >MORT? Syste'ac ;ause #nalysis TechniGue >S;#T? SeGuenally Ti'ed "vents loJng >ST"? TapRoot[ 2ncident 2nvesgaon Syste' TechniGue of Operaons Revie7 >TOR? or5 Safety #nalysis  roprietary techniGues that reGuires a license agree'ent. These t7o tables list 'ore than =1 di:erent 'ethods, but do not include a co'plete list of 'ethods. Other 'ethods are described else7here in the literature. Since DO"Es or5boo5 ;onducng #ccident 2nvesgaon >DO", +++? is a co'prehensive and 7ell( 7riWen handboo5, the descripon of accident invesgaon 'ethods starts 7ith DO"Es core analycal techniGues in secon *.. Their core analycal techniGues are$  "vents and ;ausal &actors ;harng and #nalysis  @arrier #nalysis  ;hange #nalysis  Root ;ause #nalysis &urther, so'e other 'ethods are described in secon *.=$  &ault tree analysis  "vent tree analysis  MORT >Manage'ent Oversight and Ris5 Tree? Methods for accident invesgaon =)  S;#T >Syste'ac ;ause #nalysis TechniGue?  ST" >SeGuenal Ti'ed "vents loJng?  MTO(analysis  #"@ Method  TR2OD(Delta  #cci(Map The four last 'ethods are neither listed in Table  nor Table =, but are co''only used 'ethods in di:erent industries in several "uropean countries. The readers should be a7are of that this chapter is purely descripve. #ny co''ents or assess'ents of the 'ethods are 'ade in chapter -. *. DO"Es core analycal techniGues- *.. "vents and causal factors charng >";&;? "vents and causal factors charng is a graphical display of the accidentEs chronology and is used pri'arily for co'piling and organising evidence to portray the seGuence of the accidentEs events. The events and causal factor chart is easy to develop and provides a clear depicon of the data. DO", +++?A . A Si'ilar to M"S in Table =. ;ondion #ccident event ;ondion Secondary event  "vent  Secondary event = "vent = "vent / "vent * Secondary events seGuence ri'ary events seGuence ;ondion ;ondion ;ondion "vents #ccidents ;ondions ;onnector Transfer bet7een lines 3T# 3ess than adeGuate >9udg'ent? ( #re acve >e.g. ]crane stri5es building]? ( Should be stated using one noun and one acve verb ( Should be GuanHed as 'uch as possible and 7here applicable ( Should indicate the date and 'e, 7hen they are 5no7n ( Should be derived fro' the event or events and conditons i''ediately preceding it ( #re passive >e.g. ]fog in the area]? ( Describe states or circu'stances rather than occurrences or events ( #s praccal, should be GuanHed ( Should indicate date and 'e if praccalCapplicable ( #re associated 7ith the corresponding event "nco'passes the 'ain events of the accident and those that for' the 'ain events line of the chart "nco'passes the events that are secondary or contribung events and those that for' the secondary line of the chart Secondary event seGuence ri'ary event seGuence ;ondions Sy'bols "vents resu'pve events resu'pve condions or assu'pons Methods for accident invesgaon /1 *..= @arrier analysis @arrier analysis is used to idenfy ha8ards associated 7ith an accident and the barriers that should have been in place to prevent it. # barrier is any 'eans used to control, prevent, or i'pede the ha8ard fro' reaching the target. @arrier analysis addresses$  @arriers that 7ere in place and ho7 they perfor'ed  @arriers that 7ere in place but not used  @arriers that 7ere not in place but 7ere reGuired  The barrier>s? that, if present or strengthened, 7ould prevent the sa'e or si'ilar accidents fro' occurring in the future. &igure + sho7s types of barriers that 'ay be in place to protect 7or5ers fro' ha8ards. &igure +. "a'ples on barriers to protect 7or5ers fro' ha8ards >DO", +++?) hysical barriers are usually easy to idenfy, but 'anage'ent syste' barriers 'ay be less obvious >e.g. eposure li'its?. The invesgator 'ust understand each barrierEs intended funcon and locaon, and ho7 it failed to prevent the accident. There eists di:erent 7ays in ) There eists di:erent barrier 'odels for prevenon of accidents based on the defence(in(depth principle in di:erent industries, see. e.g. =111? for prevenon of Hres and eplosions in hydrocarbon processing plants and 2NS#B(= for basic safety principles for nuclear po7er plants. Types of barriers Manage'ent barriers ( 4a8ard analysis ( hat 7as the barrierEs purposeI as the barrier in place or not in placeI Did the barrier failI as the barrier used if it 7as in placeI? Record in colu'n t7o. Step * 2denfy and consider probable causes of the barrier failure. Record in colu'n three. Step - "valuate the conseGuences of the failure in this accident. Record in colu'n four. Methods for accident invesgaon /= Table /. @arrier analysis 7or5sheet. 4a8ard$ /.= 5% electrical cable Target$ #cng pipeHWer hat 7ere the barriersI 4o7 did each barrier perfor'I hy did the barrier failI 4o7 did the barrier a:ect the accidentI "ngineering dra7ings Dra7ings 7ere inco'plete and did not idenfy electrical cable at su'p locaon "ngineering dra7ings and construcon speciHcaons 7ere not procured Dra7ings used 7ere preli'inary No as(built dra7ings 7ere used to idenfy locaon of ulity lines "istence of electrical cable un5no7n 2ndoor ecavaon per'it 2ndoor ecavaon per'it 7as not obtained ipeHWers and ulity specialist 7ere una7are of indoor ecavaon per'it reGuire'ents Opportunity to idenfy eistence of cable 'issed *../ ;hange analysis ;hange is anything that disturbs the PbalanceQ of a syste' operang as planned. ;hange is oen the source of deviaons in syste' operaons. ;hange analysis ea'ines planned or unplanned changes that caused undesired outco'es. 2n an accident invesgaon, this techniGue is used to ea'ine an accident by analysing the di:erence bet7een 7hat has occurred before or 7as epected and the actual seGuence of events. The invesgator perfor'ing the change analysis idenHes speciHc di:erences bet7een the accident\free situaon and the accident scenario. These di:erences are evaluated to deter'ine 7hether the di:erences caused or contributed to the accident. The change analysis process is described in &igure . hen conducng a change analysis, invesgators idenfy changes as 7ell as the results of those changes. The disncon is i'portant, because idenfying only the results of change 'ay not pro'pt invesgators to Methods for accident invesgaon // idenfy all causal factors of an accident. hen conducng a change analysis, it is i'portant to have a baseline situaon that the accident seGuence 'ay be co'pared to. &igure . The change analysis process. >DO", +++? Table * sho7s a si'ple change analysis 7or5sheet. The invesgators should Hrst categorise the changes according to the Guesons sho7n in the le colu'n of the 7or5sheet, i.e., deter'ine if the change pertained to, for ea'ple, a di:erence in$  hat events, condions, acvies, or eGuip'ent 7ere present in the accident situaon that 7ere not present in the baseline >accident(free, prior, or ideal? situaon >or vice versa?  hen an event or condion occurred or 7as detected in the accident situaon versus the baseline situaon  here an event or condion occurred in the accident situaon versus 7here an event or condion occurred in the baseline situaon  ho 7as involved in planning, revie7ing, authorising, perfor'ing, and supervising the 7or5 acvity in the accident versus the accident(free situaon.  4o7 the 7or5 7as 'anaged and controlled in the accident versus the accident(free situaon. To co'plete the re'ainder of the 7or5sheet, Hrst describe each event or condion of interest in the second colu'n. Then describe the related event or condion that occurred >or should have occurred? in the baseline situaon in the third colu'n. The di:erence bet7een the event and condions in the accident and the baseline situaons should Describe accident situaon Describe co'parable accident(free situaon 2nput results into events and causal factors chart #nalyse di:erences for e:ect on accident 2denfy di:erences ;o'pare Methods for accident invesgaon /* be brieFy described in the fourth colu'n. 2n the last colu'n, discuss the e:ect that each

change had on the accident. The di:erences or changes idenHed can generally be described as causal factors and should be noted on the events and causal factors chart and used in the root cause analysis. # potenal 7ea5ness of change analysis is that it does not consider the co'pounding e:ects of incre'ental change >for ea'ple, a change that 7as instuted several years earlier coupled 7ith a 'ore recent change?. To overco'e this 7ea5ness, invesgators 'ay choose 'ore than one baseline situaon against 7hich to co'pare the accident scenario. Table *. # si'ple change analysis 7or5sheet. >DO", +++? &actors #ccident situaon rior, ideal, or acciden^ree situaon Di:erence "valuaon of e:ect hat ;ondions Occurrences #cvies "Guip'ent hen Occurred 2denHed &acility status Schedule here hysical locaon "nviron'ental condions ho Sta: involved Training !ualiHcaon Supervision 4o7 ;ontrol chain 4a8ard analysis Monitoring Other Methods for accident invesgaon /- *..* "vents and causal factors analysis The events and causal factors chart 'ay also be used to deter'ine the causal factors of an accident, as illustrated in &igure =. This process is an i'portant Hrst step in later deter'ining the root causes of an accident. "vents and causal factors analysis reGuires deducve reasoning to deter'ine 7hich events andCor condions that contributed to the accident. &igure =. "vents and causal factors analysis. >DO", +++? @efore starng to analyse the events and condions noted on the chart, an invesgator 'ust Hrst ensure that the chart contains adeGuate detail. "a'ine the Hrst event that i''ediately precedes the accident. "valuate its signiHcance in the accident seGuence by as5ing$ P2f this event had not occurred, 7ould the accident have occurredIQ 2f the ans7er is yes, then the event is not signiHcant. roceed to the net event in the chart, 7or5ing bac57ards fro' the accident. 2f the ans7er is no, then deter'ine 7hether the event represented nor'al acvies 7ith the epected conseGuences. 2f the event 7as intended and had the epected outco'es, then it is not signiHcant. 4o7ever, if the event deviated fro' 7hat 7as intended or had un7anted conseGuences, then it is a signiHcant event. ;ondion ;ausal factor ;ausal factor ;ondion ;ondion "vent "vent "vent "vent 4o7 did the condions originateI hy did the syste' allo7 the condions to eistI hy did this event happenI #s5 Guesons to deter'ine causal factors >7hy, ho7, 7hat, and 7ho? "vent chain Methods for accident invesgaon /A ;arefully ea'ine the events and condions associated 7ith each signiHcant event by as5ing a series of Guesons about this event chain, such as$  hy did this event happenI  hat events and condions led to the occurrence of the eventI  hat 7ent 7rong that allo7ed the event to occurI  hy did these condions eistI  4o7 did these condions originateI  ho had the responsibility for the condionsI  #re there any relaonships bet7een 7hat 7ent 7rong in this event chain and other events or condions in the accident seGuenceI  2s the signiHcant event lin5ed to other events or condions that 'ay indicate a 'ore general or larger deHciencyI The signiHcant events, and the events and condions that allo7ed the signiHcant events to occur, are the accidentEs causal factors. *..- Root cause analysis Root cause analysis is any analysis that idenHes underlying deHciencies in a safety 'anage'ent syste' that, if corrected, 7ould prevent the sa'e and si'ilar accidents fro' occurring. Root cause analysis is a syste'ac process that uses the facts and results fro' the core analyc techniGues to deter'ine the 'ost i'portant reasons for the accident. hile the core analyc techniGues should provide ans7ers to Guesons regarding 7hat, 7hen, 7here, 7ho, and ho7, root cause analysis should resolve the Gueson 7hy. Root cause analysis reGuires a certain a'ount of 9udg'ent. # rather ehausve list of causal factors 'ust be developed prior to the applicaon of root cause analysis to ensure that Hnal root causes are accurate and co'prehensive. One 'ethod for root cause analysis described by DO" is T2"R diagra''ing. T2"R(diagra''ing is used to idenfy both the root causes of an accident and the level of line 'anage'ent that has the responsibility and authority to correct the accidentEs causal factors. The invesgators use T2"R(diagra's to hierarchically categorise the causal

factors derived fro' the events and causal factors analysis. Methods for accident invesgaon /) 3in5ages a'ong causal factors are then idenHed and possible root causes are developed. # di:erent diagra' is developed for each organisaon responsible for the 7or5 acvies associated 7ith the accident. The causal factors idenHed in the events and causal factors chart are input to the T2"R( diagra's. #ssess 7here each causal factor belong in the T2"R(diagra'. #er arranging all the causal factors, ea'ine the causal factors to deter'ine 7hether there is lin5age bet7een t7o or 'ore of the'. "valuate each of the causal factors state'ents if they are root causes of the accident. There 'ay be 'ore than one root cause of a parcular accident. &igure / sho7s an ea'ple on a T2"R(diagra'. &igure /. 2denfying the lin5ages to the root causes fro' a T2"R(diagra'. *.= Other accident invesgaon 'ethods *.=. &ault tree analysis0 &ault tree analysis is a 'ethod for deter'ining the causes of an accident >or top event?. The fault tree is a graphic 'odel that displays the various co'binaons of nor'al events, eGuip'ent failures, hu'an errors, and environ'ental factors that can result in an accident. #n ea'ple of a fault tree is sho7n in &igure *. 0 The descripon is based on 4Yyland  Rausand, ++*. Tier ;ausal &actors Tier -$ Senior 'anage'ent Tier $ or5er acons Tier =$ Supervision Tier /$ 3o7er 'anage'ent Tier *$ Middle 'anage'ent Tier 1$ Direct cause Root causes >oponal colu'n? Root cause _  Root cause _ / Root cause _ = Methods for accident invesgaon /0 &igure *. 2llustraon of a fault tree >ea'ple fro' the Vsta(accident?. # fault tree analysis 'ay be Gualitave, Guantave, or both. ossible results fro' the analysis 'ay be a lisng of the possible co'binaons of environ'ental factors, hu'an errors, nor'al events and co'ponent failures that 'ay result in a crical event in the syste' and the probability that the crical event 7ill occur during a speciHed 'e interval. The strengths of the fault tree, as a Gualitave tool is its ability to brea5 do7n an accident into root causes. The undesired event appears as the top event. This event is lin5ed to the basic failure events by logic gats and event state'ents. # gate sy'bol can have one or 'ore inputs, but only one output. # su''ary of co''on fault tree sy'bols is given in &igure -. 4Yyland and Rausand >++*? give a 'ore detailed descripon of fault tree analysis. Malfuncon of the signalling syste' 4u'an error >engine driver? 3ine secon already ]occupied] by another train SabotageC act of terros "ngine failure >runa7ay train? Or Or No signal Breen signal >green Fash? Methods for accident invesgaon /+ &igure -. &ault tree sy'bols. *.=.= "vent tree analysis+ #n event tree is used to analyse event seGuences follo7ing aer an iniang event. The event seGuence is inFuenced by either success or failure of nu'erous barriers or safety funconsCsyste's. The event seGuence leads to a set of possible conseGuences. The conseGuences 'ay be considered as acceptable or unacceptable. The event seGuence + The descripon is based on %ille'eur, ++. # #nd " " "= / # Or " " "= / The OR(gate indicates that the output event # occurs if any of the input events "i occur. The #ND(gate indicates that the output event # occurs 7hen all the input events "i occur si'ultaneously. @asic event Undeveloped event ;o''ent rectangle The basic event represents a basic eGuip'ent failure that reGuires no further develop'ent of failure causes The undeveloped event represents an event that is not ea'ined further because infor'aon is unavailable or because its conseGuences is insigniHcant The co''ent rectangle is for supple'entary infor'aon The transfer(out sy'bol indicates that the fault tree is developed further at the occurrence of the corresponding Transfer(in sy'bol Transfer(out Transfer(in 3ogic gates 2nput events Descripon of state Transfer sy'bols #ND(gate OR(gate Sy'bol Descripon Methods for accident invesgaon *1 is illustrated graphically 7here each safety syste' is 'odelled for t7o states, operaon and failure. &igure A illustrates an event tree of the situaon on RYrosbanen 9ust before the Vsta(accident. This event tree reveals the lac5 of reliable safety barriers in order to prevent train collision at RYrosbanen at that 'e. #n event tree analysis is pri'arily a proacve ris5 analysis 'ethod used to idenfy possible event

seGuences. The event tree 'ay be used to idenfy and illustrate event seGuences and also to obtain a Gualitave and Guantave representaon and assess'ent. 2n an accident invesgaon 7e 'ay illustrate the accident path as one of the possible event seGuences. This is illustrated 7ith the thic5 line in &igure A. &igure A. Si'pliHed event tree analysis of the ris5 at RYrosbanen 9ust before the Vsta( accident. *.=./ MORT1 MORT provides a syste'ac 'ethod >analyc tree? for planning, organising, and conducon a co'prehensive accident invesgaon. Through MORT analysis, invesgators idenfy deHciencies in speciHc 1 The descripon is based on ohnson .B., +01. T7o trains at the sa'e secon of the line #T; >#uto'ac Train ;ontrol? The rail trac controller detects the ha8ardous situaon Train drivers stop the train The rail trac controller alerts about the ha8ard 6es 6es 6es 6es No No No No ;ollision ;ollison avoided ;ollision ;ollision ;ollison avoided Methods for accident invesgaon * control factors and in 'anage'ent syste' factors. These factors are evaluated and analysed to idenfy the causal factors of the accident. @asically, MORT is a graphical chec5list in 7hich contains generic Guesons that invesgators aWe'pt to ans7er using available factual data. This enables invesgators to focus on potenal 5ey causal factors. The upper levels of the MORT diagra' are sho7n in &igure ). MORT reGuires etensive training to e:ecvely perfor' an in(depth analysis of co'ple accidents involving 'ulple syste's. The Hrst step of the process is to select the MORT chart for the safety progra' area of interest. The invesgators 7or5 their 7ay do7n through the tree, level by level. "vents should be coded in a speciHc colour relave to the signiHcance of the accident. #n event that is deHcient, or 3ess Than #deGuate >3T#? in MORT ter'inology is 'ar5ed red. The sy'bol is circled if suspect or coded in red if conHr'ed. #n event that is sasfactory is 'ar5ed green in the sa'e 'anner. Un5no7ns are 'ar5ed in blue, being circled inially and coloured if sucient data do not beco'e available, and an assu'pon 'ust be 'ade to connue or conclude the analysis. hen the appropriate seg'ents of the tree have been co'pleted, the path of cause and e:ect >fro' lac5 of control by 'anage'ent, to basic causes, contributory causes, and root causes? can easily be traced bac5 through the tree. The tree highlights Guite clearly 7here controls and correcve acons are needed and can be e:ecve in prevenng recurrence of the accident. Methods for accident invesgaon *= &igure ). The upper levels of the MORT(tree. "T >ro9ect "valuaon Tree? and SMORT >Safety Manage'ent and Organisaons Revie7 TechniGue? are both 'ethods based on MORT but si'pliHed and easier to use. "T and SMORT 7ill not be described further. "T is described by DO" >+++? and SMORT by +0)?. *.=.* Syste'ac ;ause #nalysis TechniGue >S;#T? The 2nternaonal 3oss ;ontrol 2nstute >23;2? developed S;#T for the support of occupaonal incident invesgaon. The 23;2 3oss ;ausaon Model is the fra'e7or5 for the S;#T syste' >see &igure 0?.  The descripon of S;#T is based on ;;S >++=? and the descripon of the 23;2('odel is based on @ird  Ber'ain >+0-?. 2n9uries, da'age, other costs, perfor'ance lost or degraded &uture undesired events  Or 2'ple'entaon 3T# #ccident #'elioraon 3T# Manage'ent syste' factos 3T# SpeciHc controls factors 3T# Oversights and o'issions #ssu'ed ris5s Ris5 assess'ent syste' 3T# Ris5  Ris5 = Ris5 / Ris5 n #nd Or Or olicy 3T# T SCM S M S# S#= M# M#/ M#= Or hat happenedI hyI # @ ; D R # Dra7ing brea5. Transfer to secon of tree indicated by sy'bol idenHcaon leWer(nu'ber Methods for accident invesgaon */ &igure 0. The 23;2 3oss ;ausaon Model >@ird and Ber'ain, +0-?. The result of an accident is loss, e.g. har' to people, properes, products or the environ'ent. The incident >the contact bet7een the source of energy and the Pvic'Q? is the event that precedes the loss. The i''ediate causes of an accident are the circu'stances that i''ediately precede the contact. They usually can be seen or sensed. &reGuently they are called unsafe acts or unsafe condions, but in the 23;2('odel the ter's substandard acts >or pracces? and substandard condions are used. Substandard acts and condions are listed in &igure +.

&igure +. Substandard acts and condions in the 23;2('odel. @asic causes are the diseases or real causes behind the sy'pto's, the reasons 7hy the substandard acts and condions occurred. @asic causes help eplain 7hy people perfor' substandard pracces and 3ac5 of control 2ncident 3oss 2''ediate causes @asic causes 2nadeGuate$ rogra' rogra' standards ;o'pliance to standards ersonal factors ob factors eople roperty roduct "nviron'ent Service ;ontact 7ith energy, substance or people Substandard acts Substandard condions Substandard praccesCacts Substandard condions . Operang eGuip'ent 7ithout authority =. &ailure to 7arn /. &ailure to secure *. Operang at i'proper speed -. Ma5ing safety devices inoperable A. Re'oving safety devices ). Using defecve eGuip'ent 0. Using eGuip'ent i'properly +. &ailing to use personal protecve eGuip'ent 1. 2'proper loading . 2'proper place'ent =. 2'proper liing /. 2'proper posion for tas5 *. Servicing eGuip'net in operaon -. 4orseplay A. Under inFuence of alcoholCdrugs . 2nadeGuate guards or barriers =. 2nadeGuate or i'proper protecve eGuip'ent /. Defecve tools, eGuip'ent or 'aterials *. ;ongeson or restricted acon -. 2nadeGuate 7arning syste' A. &ire and eplosion ha8ards ). oor house5eeping, disorderly 7or5place 0. 4a8ardous environ'ental condions +. Noise eposures 1. Radiaon eposures . 4igh or lo7 te'perature eposures =. 2nadeGuate or ecessive illu'inaon /. 2nadeGuate venlaon Methods for accident invesgaon ** 7hy substandard condions eists. #n overvie7 of personal and 9ob factors are given in &igure =1. &igure =1. ersonal and 9ob factors in the 23;2('odel. There are three reasons for lac5 of control$ . 2nadeGuate progra' =. 2nadeGuate progra' standards and /. 2nadeGuate co'pliance 7ith standards &igure = sho7s the ele'ents that should be in place in a safety progra'. The ele'ents are based on research and eperience fro' successful safety progra's in di:erent co'panies. &igure =. "le'ents in a safety progra' in the 23;2('odel. The Syste'ac ;ause #nalysis TechniGue is a tool to aid an invesgaon and evaluaon of incidents through the applicaon of a S;#T chart. The chart acts as a chec5list or reference to ensure that an invesgaon has loo5ed at all facets of an incident. There are Hve ersonal factors ob factors . 2nadeGuate capability ( hysicalCphysiological ( MentalCpsychological =. 3ac5 of 5no7ledge /. 3ac5 of s5ill *. Stress ( hysicalCphysiological ( MentalCpsychologica -. 2'proper 'ovaon . 2nadeGuate leadership andCor supervision =. 2nadeGuate engineering /. 2nadeGuate purchasing *. 2nadeGuate 'aintenance -. 2nadeGuate tools, eGuip'ent, 'aterials A. 2nadeGuate 7or5 standards ). ear and tear 0. #buse or 'isuse "le'ents in a safety progra' . 3eadership and ad'inistraon =. Manage'ent training /. lanned inspecon *. Tas5 analysis and procedures -. #ccidentCincident invesgaon A. Tas5 observaons ). "'ergency preparedness 0. Organisaonal rules +. #ccidentCincident analysis 1. "'ployee training . ersonal protecve eGuip'ent =. 4ealth control /. rogra' evaluaon syste' *. "ngineering controls -. ersonal co''unicaons A. Broup 'eengs ). Beneral pro'oon 0. 4iring and place'ent +. urchasing controls =1. O:(the(9ob safety Methods for accident invesgaon *- bloc5s on a S;#T chart. "ach bloc5 corresponds to a bloc5 of the loss causaon 'odel. 4ence, the Hrst bloc5 contains space to 7rite a descripon of the incident. The second bloc5 lists the 'ost co''on categories of contact that could have led to the incident under invesgaon. The third bloc5 lists the 'ost co''on i''ediate causes, 7hile the fourth bloc5 lists co''on basic causes. &inally, the boWo' bloc5 lists acvies generally accepted as i'portant for a successful loss control progra'. The techniGue is easy to apply and is supported by a training 'anual. The S;#T see's to correspond to the S6N"RB2 tool for accident registraon used in Nor7ay. #t least, the accident causaon 'odels used in S;#T and S6N"RB2 are eGuivalent. *.=.- ST" >SeGuenal 'ed events ploJng?= The ST"('ethod 7as developed by 4endric5 and @enner >+0)?. They propose a syste'ac process for accident invesgaon based on 'ul(linear events seGuences and a process vie7 of the accident pheno'ena. ST" builds on

four concepts$ . Neither the accident nor its invesgaon is a single linear chain or seGuence of events. Rather, several acvies ta5e place at the sa'e 'e. =. The event @uilding @loc5 for'at for data is used to develop the accident descripon in a 7or5sheet. # building bloc5 describes one event, i.e. one actor perfor'ing one acon. /. "vents Fo7 logically during a process. #rro7s in the ST" 7or5sheet illustrate the Fo7. *. @oth producve and accident processes are si'ilar and can be understood using si'ilar invesgaon procedures. They both involve actors and acons, and both are capable of being repeated once they are understood. ith the process concept, a speciHc accident begins 7ith the acon that started the transfor'aon fro' the described process to an = The descripon is based on 4endric5  @enner, +0). Methods for accident invesgaon *A accident process, and ends 7ith the last connected har'ful event of that accident process. The ST"(7or5sheet provides a syste'ac 7ay to organise the building bloc5s into a co'prehensive, 'ul(linear descripon of the accident process. The ST"( 7or5sheet is si'ply a 'atri, 7ith ro7s and colu'ns. There is one ro7 in the 7or5sheet for each actor. The colu'ns are labelled di:erently, 7ith 'ar5s or nu'bers along a 'e line across the top of the 7or5sheet, as sho7n in &igure ==. The 'e scale does not need to be dra7n on a linear scale, the 'ain point of the 'e line is to 5eep events in order, i.e., ho7 they relate to each other in ter's of 'e. &igure ==. ST"(7or5sheet. #n event is one actor perfor'ing one acon. #n actor is a person or an ite' that directly inFuences the Fo7 or events constung the accident process. #ctors can be involved in t7o types of changes, adapve changes or iniang changes. They can either change reacvely to sustain dyna'ic balance or they can introduce changes to 7hich other actors 'ust adapt. #n acon is so'ething done by the actor. 2t 'ay be physical and observable, or it 'ay be 'ental if the actor is a person. #n acon is so'ething that the actor does and 'ust be stated in the acve voice. The ST" 7or5sheet provides a syste'ac 7ay to organise the building bloc5s >or events? into a co'prehensive, 'ul(linear descripon of the accident process. &igure =/ sho7s an ea'ple on a #ctor # #ctor D #ctor ; #ctor @ T1 Ti'e "tc. Methods for accident invesgaon *) ST"(diagra' of an accident 7here a stone bloc5 falls o: a truc5 and hits a car/. &igure =/. #n ea'ple on a si'ple ST"(diagra' for a car accident. The ST"(diagra' in &igure =/ also sho7s the use of arro7s to lin5 tested relaonships a'ong events in the accident chain. #n arro7 convenon is used to sho7 precedeCfollo7 and logical relaons bet7een t7o or 'ore events. hen an earlier acon is necessary for a laWer to occur, an arro7 should be dra7n fro' the preceding event to the resultant event. The thought process for idenfying the lin5s bet7een events is related to the change of state concepts underlying ST" 'ethods. &or each event in the 7or5sheet, the invesgator as5s, P#re the preceding acons sucient to iniate this acons >or event? or 7ere other acons necessaryIQ Try to visuali8e the actors and acons in a P'ental 'ovieQ in order to develop the lin5s. So'e'es it is i'portant to deter'ine 7hat happened during a gap or 'e interval for 7hich 7e cannot gather any speciHc evidence. "ach re'aining gap in the 7or5sheet represents a gap in the understanding of the accident. @ac5ST" is a techniGue by 7hich you reason your 7ay bac57ards fro' the event on the right side of the 7or5sheet gap / The ST"(diagra' is based on a descripon of the accident in a ne7spaper arcle. T1 Ti'e ;ar ;ar driver Drap Stone bloc5 Truc5 Truc5 driver loads stone on truc5 Truc5 driver drives truc5 fro' # to @ Truc5 drives fro' # to @ ;ar driver dies Truc5 driver fastens the stone bloc5 ;ar drive fro' @ to # ;ar driver tries to avoid to hit the stone The car hits the stone bloc5 Stone falls o: the truc5 Drap fails The car ]collapses] >collision da'aged? ;ar driver starts the car 3egend Truc5 driver Drap fails #ctor "vent lin5 #ctor Truc5 driver ;ar driver observes the stone ;ar driver stri5es Methods for accident invesgaon *0 to7ard the event on the le side of the gap. The @ac5ST" procedure consists of as5ing a series of Phat could have led to thatIQ Guesons and 7or5ing bac57ard through the pyra'id 7ith the ans7ers. Ma5e tentave event building bloc5s for each

event that ans7ers the Gueson. hen doing a @ac5ST", it is not unco''on to idenfy 'ore than one possible path7ay bet7een the le and right events at the gap. This tells that there 'ay be 'ore than one 7ay the accident process could progress and 'ay led to develop'ent of hypothesis in 7hich should be further ea'ined. The ST"(procedure also includes so'e rigorous technical truthtesng procedures, the ro7 test, the colu'n test, and the necessaryand(sucient test. The ro7 >or hori8ontal? test tells you if you need 'ore building bloc5s for any individual actor listed along the le side of the 7or5sheet. 2t also tells you if you have bro5en each actor do7n suciently. The colu'n >or vercal? test chec5s the seGuence of events by pairing the ne7 event 7ith the acons of other actors. To pass the colu'n test, the event building bloc5 being tested 'ust have occurred  #er all the event in all the colu'ns to the le of that event,  @efore all the events in all colu'ns to the right of that event, and  #t the sa'e 'e as all the events in the sa'e colu'n. The ro7 test and the colu'n test are illustrated in &igure =*. &igure =*. or5sheet ro7 test and colu'n test. #ctor # #ctor D #ctor ; #ctor @ T1 Ti'e "tc. ;olu'ns Ro7s Ro7 tests 2s ro7 co'pleteI ;olu'n tests 2s event seGuenced Ocollision da'aged? ;ar driver starts the car Truc5 driver ;ar driver observes the stone ;ar driver bra5es @u'by road due to lac5 of 'aintenance 2naWenve car driver A ) 0 Narro7 road oor bra5es + 1 3ac5 of airbag Methods for accident invesgaon - invesgaon. The 'ethod is based on 4"S >4u'an erfor'ance "nhance'ent Syste'? 7hich is 'enoned in Table =, but not described further in this report. The MTO( analysis is based on three 'ethods$ . Structured analysis by use of an event( and cause(diagra'). =. ;hange analysis by describing ho7 events have deviated fro' earlier events or co''on pracce0. /. @arrier analysis by idenfying technological and ad'inistrave barriers in 7hich have failed or are 'issing+. &igure =A illustrates the MTO(analysis 7or5sheet. The Hrst step in an MTO(analysis is to develop the event seGuence longitudinally and illustrate the event seGuence in a bloc5 diagra'. 2denfy possible technical and hu'an causes of each event and dra7 these vercally to each event in the diagra'. &urther, analyse 7hich technical, hu'an or organisaonal barriers that have failed or 7as 'issing during the accident progress. 2llustrate all 'issing or failed barriers belo7 the events in the diagra'. #ssess 7hich deviaons or changes in 7hich di:er the accident progress fro' the nor'al situaon. These changes are also illustrated in the diagra' >see &igure =A?. The basic Guesons in the analysis are$  hat 'ay have prevented the connuaon of the accident seGuenceI  hat 'ay the organisaon have done in the past in order to prevent the accidentI The last i'portant step in the MTO( analysis is to idenfy and present reco''endaons. The reco''endaons should be as realisc and speciHc as possible, and 'ight be technical, hu'an or organisaonal. ) See subsecon *... 0 See subsecon *../. + See subsecon *..=. Methods for accident invesgaon -= &igure =A. MTO(analysis 7or5sheet. # chec5list for idenHcaon of failure causes >Pfelorsa5erQ? is also part of the MTO( 'ethodology >@ento, +++?. The chec5list contains the follo7ing factors$ . Organisaon =. or5 organisaon /. or5 pracce *. Manage'ent of 7or5 -. ;hange procedures A. "rgono'ic C deHciencies in the technology ). ;o''unicaon 0. 2nstruconsCprocedures +. "ducaonCco'petence 1. or5 environ'ent ;hange analysis "vents and causes chart @arrier analysis  Nor'al Deviaon =  >;hain of events? >;auses? Methods for accident invesgaon -/ &or each of these failure causes, there is a detailed chec5list for basic or funda'ental causes >Pgrundorsa5erQ?. "a'ples on basic causes for the failure cause 7or5 pracce are$  Deviaon fro' 7or5 instrucon  oor preparaon or planning  3ac5 of self inspecon  Use of 7rong eGuip'ent  rong use of eGuip'ent *.=.) #ccident #nalysis and @arrier &uncon >#"@? Method=1 The #ccident "voluon and @arrier &uncon >#"@? 'odel provides a 'ethod for analysis of incidents and accidents that 'odels the evoluon to7ards an incidentCaccident as a series of interacons bet7een hu'an and technical syste's. The interacon consists of failures, 'alfuncons or errors that could lead to or have resulted in an accident. The 'ethod forces analysts to integrate hu'an and technical syste's si'ultaneously 7hen perfor'ing an accident analysis starng 7ith the si'ple Fo7 chart techniGue of the 'ethod. The Fo7 chart inially consists of e'pty boes in t7o parallel colu'ns, one for the hu'an syste's and one for the technical syste's. &igure =) provides

an illustraon of this diagra'. During the analysis these error boes are idenHed as the failures, 'alfuncons or errors that constute the accident evoluon. 2n general, the seGuence of error boes in the diagra' follo7s the 'e order of events. @et7een each pair of successive error boes there is a possibility to arrest the evoluon to7ards an incidentCaccident. @arrier funcon syste's >e.g. co'puter progra's? that are acvated can arrest the evoluon through e:ecve barrier funcons >e.g. the co'puter 'a5ing an incorrect hu'an intervenon 'odelled in the net error bo i'possible through bloc5ing a control?. &actors that have an inFuence on hu'an perfor'ance have been called perfor'ance shaping factors >by S7ain and BuW'an, +0/?. "a'ples of such factors are alcohol, lac5 of sleep and stress. 2n applicaon of the #"@ 'odel those factors are included in the Fo7 =1 The descripon is based on Svensson, =111. Methods for accident invesgaon -* diagra' only as S&s and they are analysed aer the diagra' has been co'pleted. S&s are included in the Fo7 diagra' in cases 7here it is possible that the factor could have contributed to one or 'ore hu'an error events. &actors such as alcohol and age are 'odelled as S&s, but never as hu'an error events or failing barrier funcons. Organisaonal factors 'ay be integrated as a barrier funcon 7ith failing or inadeGuate barrier funcons. Organisaonal factors should al7ays be treated in a special 7ay in an #"@ analysis because they include both hu'an and technical syste's. &igure =). 2llustraon of an #"@ analysis. #n #"@ analysis consists of t7o 'ain phases. The Hrst phase is to 'odel the accident evoluon in a Fo7 diagra'. 2t is i'portant to re'e'ber that #"@ only 'odels errors and that it is not an event seGuence 'ethod. #rro7s lin5 the error event boes together in order to sho7 the evoluon. The course of events is described in an approi'ate chronological order. 2t is not allo7ed to let 'ore than one arro7 lead to an error bo or to have 'ore than one arro7 going fro' a bo. The second phase consists of the barrier funcon analysis. 2n this phase, the barrier funcons are idenHed >ine:ecve andCor non eistent?. # barrier funcon represents a funcon that can arrest the accident evoluon so that the net event in the chain 7ill not be realised. # barrier funcon is al7ays idenHed in relaon to the syste's it protects, protected or could have protected. @arrier funcon syste's are the syste's perfor'ing the barrier funcons. @arrier funcon syste's can be an operator, an instrucon, a physical 4u'an error event  Technical error event  4u'an error event / 4u'an error event = #ccident C incident Technical error event = 4u'an factors syste' Technical syste' ;o''ents S& erfor'ance shaping factors &ailing orCand possible barrier funcon 3egend "rror event bo #ccidentCincident #rro7s describing the accident evoluon ossible barrier funcons ":ecve barrier funcon S& erfor'ing shaping factors Methods for accident invesgaon -- separaon, an e'ergency control syste', other safety(related syste's, etc. The sa'e barrier funcon can be perfor'ed by di:erent barrier funcon syste's. ;orrespondingly, a barrier funcon syste' 'ay perfor' di:erent barrier funcons. #n i'portant purpose of the #"@(analysis is to idenfy bro5en barrier funcons, the reasons for 7hy there 7ere no barrier funcons or 7hy the eisng ones failed, and to suggest i'prove'ents. @arrier funcons belong to one of the three 'ain categories$  2ne:ecve barrier funcons \ barrier funcons that 7ere ine:ecve in the sense that they did not prevent the develop'ent to7ard an accident  Non(eisng barrier funcons \ barrier funcons that, if present, 7ould have stopped the accident evoluon.  ":ecve barrier funcons \ barrier funcons that actually prevented the progress to7ard an accident. 2f a parcular accident should happen, it is necessary that all barrier funcons in the seGuence are bro5en and ine:ecve. The ob9ecve of an #"@( analysis is to understand 7hy a nu'ber of barrier funcons failed, and ho7 they could be reinforced or supported by other barrier funcons. &ro' this perspecve, idenHcaon of a root(cause of an accident is 'eaningless. The starng point of the analysis cannot be regarded as the root cause because the re'oval of any of all the other errors in the accident evoluon 7ould also eli'inate the accident. 2t is

so'e'es dicult to 5no7 if an error should be 'odelled as an error or as a failing barrier funcon. #s a rule of thu'b, 7hen uncertain the analysts should choose a bo and not a barrier funcon representaon in the inial #"@(analysis. The barrier funcon analysis phase 'ay be used for 'odelling of subsyste's interacons that cannot be represented seGuenally in #"@. #ll barriers funcon failures, incidents and accidents ta5e place in 'an \ technology \ organisaons contets. Therefore, an #"@( analysis also includes issues about the contet in 7hich the accident too5 place. Therefore, the follo7ing Guesons have to be ans7ered$ Methods for accident invesgaon -A . To increase safety, ho7 is it possible to change the organisaon, in 7hich the failure or accident too5 placeI =. To increase safety, ho7 is it possible to change the technical syste's contet, in 7hich the failure or accident too5 placeI 2t is i'portant to bear in 'ind that 7hen changes are 'ade in the organisaonal and technical syste's at the contet level far reaching e:ects 'ay be aWained. *.=.0 TR2OD= The 7hole research into the TR2OD concept started in +00 7hen a study that 7as contained in the report PTR2OD, # principled basis for accident prevenonQ >Reason et al, +00? 7as presented to Shell 2nternaonale etroleu' Maatschappi9, "ploraon and roducon. The idea behind TR2OD is that organisaonal failures are the 'ain factors in accident causaon. These factors are 'ore PlatentQ and, 7hen contribung to an accident, are al7ays follo7ed by a nu'ber of technical and hu'an errors. The co'plete TR2OD( 'odel== is illustrated in &igure =0. &igure =0. The co'plete TR2OD 'odel. Substandard acts and situaons do not 9ust occur. They are generated by 'echanis's acng in organisaons, regardless 7hether there has been an accident or not. Oen these 'echanis's result fro' decisions = This descripon is based on Broene7eg, ++0. == The TR2OD('odel described here 'ight be di:erent fro' previously published 'odels based on the TR2OD theory, but this 'odel is fully co'pable 7ith the 'ost resent version of the accident invesgaon tool TR2OD @eta described later in this chapter. Decision 'a5ers Substandard acts sychological precursors Operaonal disturbance ;onseGuences #ccident @reached barriers @reached barriers 3atent failures 1 @R&s 3atent failures @R& Defences Methods for accident invesgaon -) ta5en at high level in the organisaon. These underlying 'echanis's are called @asic Ris5 &actors=/ >@S&s?. These @S&s 'ay generate various psychological precursors in 7hich 'ay lead to substandard acts and situaons. "a'ples on psychological precursors of slips, lapses and violaons are 'e pressure, being poorly 'ovated or depressed. #ccording to this 'odel, eli'inang the latent failures categori8ed in @R&s or reducing their i'pact 7ill prevent psychological precursors, substandard acts and the operaonal disturbances. &urther'ore, this 7ill result in prevenon of accidents. The idenHed @R&s cover hu'an, organisaonal and technical proble's. The di:erent @asic Ris5 &actors are deHned in Table -. Ten of these @R&s leading to the Poperaonal disturbanceQ >the PprevenveQ @R&s?, and one @R& is ai'ed at controlling the conseGuences once the operaonal disturbance has occurred >the P'igaonQ @R&?. There are Hve generic prevenon @R&s >A \ 1 in Table -? and Hve speciHc @R&s > \ - in Table -?. The speciHc @R&s relate to latent failures that are speciHc for the operaons to be invesgated >e.g. the reGuire'ents for Tools and "Guip'ent are Guite di:erent in a oil drilling environ'ent co'pared to an intensive care 7ard in a hospital?. These  @R&s have been idenHed as a result of brainstor'ing, a study of audit reports, accident scenarios, a theorecal study, and a study on o:shore pla^or's. The division is deHnive and has sho7n to be valid for all industrial applicaons. =/ These 'echanis's 7ere inially called Beneral &ailure Types >B&Ts?. Methods for accident invesgaon -0 Table -. The deHnions of the basic ris5 factors >@R&s? in TR2OD. No @asic Ris5 &actor #bbr. DeHnion  Design D" "rgono'ically poor design of tools or eGuip'ent >user( unfriendly? = Tools and eGuip'ent T" oor Guality, condion, suitability or availability of 'aterials, tools, eGuip'ent and co'ponents / Maintenance 'anage'ent MM No or inadeGuate perfor'ance of

'aintenance tas5s and repairs * 4ouse5eeping 4< No or insucient aWenon given to 5eeping the 7or5 Foor clean or died up - "rror enforcing condions "; Unsuitable physical perfor'ance of 'aintenance tas5s and repairs A rocedures R 2nsucient Guality or availability of procedures, guidelines, instrucons and 'anuals >speciHcaons, Ppaper7or5Q, use in pracce? ) Training TR No or insucient co'petence or eperience a'ong e'ployees >not suciently suitedCinadeGuately trained? 0 ;o''unicaon ;O No or ine:ecve co''unicaon bet7een the various sites, depart'ents or e'ployees of a co'pany or 7ith the ocial bodies + 2nco'pable goals 2B The situaon in 7hich e'ployees 'ust choose bet7een op'al 7or5ing 'ethods according to the established rules on one hand, and the pursuit of producon, Hnancial, polical, social or individual goals on the other 1 Organisaon OR Shortco'ings in the organisaonEs structure, organisaonEs philosophy, organisaonal processes or 'anage'ent strategies, resulng in inadeGuate or ine:ecve 'anage'ent of the co'pany  Defences D& No or insucient protecon of people, 'aterial and environ'ent against the conseGuences of the operaonal disturbances TR2OD @eta The TR2OD @eta(tool is a co'puter(based instru'ent that provides the user 7ith a tree(li5e overvie7 of the accident that 7as invesgated. 2t is a 'enu driven tool that 7ill guide the invesgator through the process of 'a5ing an electronic representaon of the accident. Methods for accident invesgaon -+ The @"T#(tool 'erges t7o di:erent 'odels, the 4"M >The 4a8ard and ":ects Manage'ent rocess? 'odel and the TR2OD 'odel. The 'erge has resulted in an incident causaon 'odel that di:ers conceptually fro' the original TR2OD 'odel. The 4"M 'odel is presented in &igure =+. &igure =+. P#ccident 'echanis'Q according to 4"M. The TR2OD @eta accident causaon 'odel is presented in &igure /1. This string is used to idenfy the causes that lead to the breaching of the controls and defences presented in the 4"M 'odel. &igure /1. TR2OD @eta #ccident ;ausaon Model. #lthough the 'odel presented in &igure /1 loo5s li5e the original TR2OD 'odel, its co'ponents and assu'pons are di:erent. 2n the @eta('odel the defences and controls are directly lin5ed to unsafe acts, precondions and latent failures. Unsafe acts describe ho7 the barriers 7ere breached and the latent failures 7hy the barriers 7ere breached. #n ea'ple of a TR2OD @eta accident analysis is sho7n in &igure /. 4a8ard #ccidentC event %ic' or target &ailed control &ailed defence #ccident &ailed controls or defences 3atent failure>s? recondion>s? #cve failure>s? Methods for accident invesgaon A1 &igure /. "a'ple on a TR2OD @eta analysis. The ne7 7ay of invesgang accidents >see &igure /=? is Guite di:erent fro' the convenonal ones. No research is done to idenfy all the contribung substandard acts or clusters of substandard acts, the target for invesgaon is to Hnd out 7hether any of the @asic Ris5 &actors are acng. hen the @R&s have been idenHed, their i'pact can be decreased or even be eli'inated. The real source of proble's is tac5led instead of the sy'pto's. 4a8ard$ ointed table corner "'ployee hits table. 2n9ured 5nee %ic' oor ha8ard register oor ha8ard register Missing control Rounded or rubber corners Missing control #udit for obstacles Missing defence the targets of control?, and the up7ard Fo7 of state infor'aon >the 'easure'ents of control?. Decision recondion Order &uncon lan Decision Order 2ndirect conseGuence Tas5 or #con Tas5 or acon Direct conseGuence ;onseGuence recondion evaluated no further riories Syste' level . Bovern'ent. olicy  budgeng =. Regulatory bodies and #ssociaons /. 3ocal area govern'ent ;o'pany 'anage'ent lanning  budgeng ;rical event 3oss of control or loss of contain'ent Direct conseGuence *. Technical  operaonal 'anage'ent -. hysical processes  #ctor acvites A. "Guip'ent  surroundings 0  -  Reference to annotaons 2nFuence Methods for accident invesgaon A- &igure /-. rincipal illustraon of an #ctorMap. #ctor Syste' level . Bovern'ent =. Regulatory bodies /. Regional  3ocal govern'ent ;o'pany 'anage'ent *. Technical  operaonal 'anage'ent #ssociaons -. Operators #ctor #ctor #ctor #ctor #ctor #ctors #ctor #ctor #ctor #ctor #ctor #ctor #ctor #ctor #ctor #ctor Methods for accident invesgaon A) - Discussion and conclusion -. Discussion ithin the Held of accident invesgaon, there are no co''on agree'ent of deHnions of concepts, it tend to be a liWle confusion of ideas. "specially the noon of cause has been discussed. hile so'e invesgators focus on causal factors >e.g. DO", ++)?, others focus on deter'ining factors >e.g. e.g. 4op5ins, =111?, acve failures and latent condions >e.g. Reason, ++)? or safety proble's >4endric5  @enner, +0)?. @R&? >Pevent triosQ?. The third colu'n covers the level of scope of the di:erent analysis 'ethods. The levels correspond to the di:erent levels of the sociotechnical syste' involved in ris5 'anage'ent illustrated in &igure *. The di:erent levels are$ . The 7or5 and technological syste' =. The sta: level /. The 'anage'ent level *. The co'pany level -. The regulators and associaons level A. The Bovern'ent level #s sho7n in Table A, the scope of 'ost of the 'ethods is li'ited to level  \ *. #lthough ST" 7as originally developed to cover level  \ *, eperience fro' S2NT"&Es accident invesgaons sho7s that the Methods for accident invesgaon A+ 'ethod also 'ay be used to analyse events inFuenced by the regulators and the Bovern'ent. 2n addion to ST", only #cci(Map put focus on level - and A. This 'eans that invesgators focusing on the Bovern'ent and the regulators in their accident invesgaon to a great etend need to base their analysis on eperience and praccal 9udge'ent 'ore than on results fro' for'al analysis 'ethods. The fourth colu'n states 7hether the 'ethods are a pri'ary 'ethod or a secondary 'ethod. ri'ary 'ethods are stand(alone techniGues 7hile secondary 'ethods provide special input as supple'ent to other 'ethods. "vents and causal factors charng, ST", MTO( analysis, TR2OD and #cci('ap are all pri'ary 'ethods. The fault tree analysis and event tree analysis 'ight be both pri'ary and secondary 'ethods. The other 'ethods are secondary 'ethods. 2n the Hh colu'n the di:erent 'ethods are categori8ed as deducve, inducve, 'orphological or non(syste' oriented. &ault tree analysis and MORT are deducve 'ethods 7hile event three analysis is an inducve 'ethod. #cci('ap 'ight be both inducve and deducve. The #"@('ethod is characteri8ed as 'orphological, 7hile the other 'ethods are non(syste' oriented. 2n the sith colu'n the 'ethods are lin5ed to di:erent types of accident 'odels in 7hich have inFuenced the 'ethods. The follo7ing accident 'odels are used$ # ;ausal(seGuence 'odel @ rocess 'odel ; "nergy 'odel D 3ogical tree 'odel " S4"('anage'ent 'odels Root cause analysis, S;#T and TR2OD are based on causal(seGuence 'odels. "vents and causal charng, change analysis, events and causal factors analysis, ST", MTO( analysis and #"@('ethod are based on process 'odels. The barrier analysis is based on the energy 'odel. &ault tree analysis, event tree analysis and MORT are based on logical tree 'odels. MORT and S;#T are also based on S4"('anage'ent 'odels. The #cci('ap is based on a co'binaon of a causal( seGuence 'odel, a process 'odel and a logical tree 'odel. Methods for accident invesgaon )1 2n the last colu'n, there is 'ade an assess'ent of the need of educaon and training in order to use the 'ethods. The ter's P"pertQ, PSpecialistQ and PNoviceQ are used in the table. "pert indicates that there is need of for'al educaon and training before people are able to use the 'ethods in a proper 7ay. So'e eperience is also beneHcial. &ault tree analysis, MORT and #cci('ap enter into this category.

Novice indicates that people are able to use the 'ethods aer and orientaon of the 'ethods 7ithout hands(on training or eperience. "vents and causal factors charng, barrier analysis, change analysis and ST" enter into this category. Specialist is so'e7here bet7een epert and novice and events and causal factors analysis, root cause analysis, event tree analysis, S;#T, MTO(analysis, #"@'ethod and TR2OD enter into this category. Methods for accident invesgaon ) Table A. ;haracteriscs of di:erent accident invesgaon 'ethods. Method #ccident seGuence 3evels of analysis ri'ary C secondary #nalycal approach #ccident 'odel Training need "vents and causal factors charng 6es (* ri'ary Non(syste' oriented @ Novice @arrier analysis No (= Secondary Non(syste' oriented ; Novice ;hange analysis No (* Secondary Non(syste' oriented @ Novice "vents and causal factors analysis (* Secondary Non(syste' oriented @ Specialist Root cause analysis No (* Secondary Non(syste' oriented # Specialist &ault tree analysis No (= ri'aryC Secondary Deducve D "pert "vent Tree analysis No (/ ri'aryC Secondary 2nducve D Specialist MORT No =(* Secondary Deducve D C " "pert S;#T No (* Secondary Non(syste' oriented # C " Specialist ST" 6es (A ri'ary Non(syste' oriented @ Novice MTOanalysis 6es (* ri'ary Non(syste' oriented @ SpecialistC epert #"@'ethod No (/ Secondary Morphological @ Specialist TR2OD 6es (* ri'ary Non(syste' oriented # Specialist #cci(Map No (A ri'ary Deducve  inducve # C @ C D "pert -.= ;onclusion Ma9or accidents al'ost never result fro' one single cause, 'ost accidents involve 'ulple, interrelated causal factors. #ll actors or decision( 'a5ers inFuencing the nor'al 7or5 process 'ight also Methods for accident invesgaon )= inFuence accident scenarios, either directly or indirectly. This co'pleity should also reFect the accident invesgaon process. The ai' of accident invesgaons should be to idenfy the event seGuences and all >causal? factors inFuencing the accident scenario in order to be able to suggest ris5 reducing 'easures in 7hich 'ay prevent future accidents. This 'eans that all 5ind of actors, fro' technical syste's and front(line operators to regulators and the Bovern'ent need to be analysed. Oen, accident invesgaons involve using of a set of accident invesgaon 'ethods. "ach 'ethod 'ight have di:erent purposes and 'ay be a liWle part of the total invesgaon process. Re'e'ber, every piece of a pu88le is as i'portant as the others. Braphical illustraons of the event seGuence are useful during the invesgaon process because it provides an e:ecve visual aid that su''aries 5ey infor'aon and provide a structured 'ethod for collecng, organising and integrang collected evidence to facilitate co''unicaon a'ong the invesgators. Braphical illustraons also help idenfying infor'aon gaps. During the invesgaon process di:erent 'ethods should be used in order to analyse arising proble' areas. #'ong the 'ul(disciplinary invesgaon tea', there should be at least one 'e'ber having good 5no7ledge about the di:erent accident invesgaon 'ethods, being able to choose the proper 'ethods for analysing the di:erent proble's. ust li5e the 'echanicians have to choose the right tool on order to repair a technical syste', an accident invesgator has to choose proper 'ethods analysing di:erent proble' areas. Methods for accident invesgaon )/ A References #ndersson R.  Menc5el "., ++-. On the prevenon of accidents and in9uries. # co'parave analysis of conceptual fra'e7or5s. #ccident #nalysis and revenon, %ol. =), No. A, )-) \ )A0. #rbeids'il9Ysenteret, =11. %eiledning i uly55esgrans5ing, #rbeids'il9Yforlaget, =11 @ento, (., +++. MTO(analys av hndelsesrapporter, OD( 11(= @ird, &.". r  Ber'ain, B.3., +0-. raccal 3oss ;ontrol 3eadership. 2S@N 1(001A(1-*(+, 2nternaonal 3oss ;ontrol 2nstute, Beorgia, US#. ;;S, ++=. Buidelines for 2nvesgang ;he'ical rocess 2ncidents. 2S@N 1(0A+(1---(, ;enter for ;he'ical rocess Safety of the #'erican 2nstute of ;he'ical "ngineers, ++=. DO", ++). 2'ple'entaon Buide &or Use ith DO" Order ==-.#, #ccident 2nvesgaons, DO" B ==-.#( Nove'ber =A, ++)CRev. , U.S. Depart'ent of "nergy, ashington D.;, US#. DO", +++. ;onducng #ccident 2nvesgaons DO" or5boo5, Revision =, May , +++, U.S.

Depart'ent of "nergy, ashington D.;, US#. &erry T.S., +00. Modern accident invesgaon and analysis >=nd ed.?. 2S@N 1.*)(A=*0(1, iley 2nterscience publicaon, United States. Bil9e, N. og Bri'en, 4., ++/. Sa'funnsvitens5apenes forutsetninger 2nnfYring i sa'funnsvitens5apenes vitens5apsHlosoH, Universitetsforlaget, Oslo. Broene7eg, ., ++0. ;ontrolling the controllable The 'anage'ent of safety. &ourth edion. DSO ress, 3eiden University, The Netherlands, ++0. 4ale #, ilpert @, &reitag M, ++). #er the event ( fro' accident to organisaonal learning. 2S@N 1 10 1*/1)*1, erga'on, ++). 4endric5
View more...

Comments

Copyright ©2017 KUPDF Inc.
SUPPORT KUPDF