Safety Assessment of High-Risk Operations in Hydroelectric ...

Hindawi Publishing CorporationMathematical Problems in EngineeringVolume 2013 Article ID 530198 12 pageshttpdxdoiorg1011552013530198

Research ArticleSafety Assessment of High-Risk Operations inHydroelectric-Project Based on Accidents AnalysisSEM and ANP

Jian-Lan Zhou1 Bai Zhe-Hua1 and Zhi-Yu Sun2

1 Key Laboratory ofMinistry of Education for Image Processing and Intelligent Control Department of Systems Science and EngineeringHuazhong University of Science amp Technology Wuhan Hubei 430074 China

2Departments of Science Technology amp Environmental Protection China Three Gorges Project Corporation Beijing 100038 China

Correspondence should be addressed to Jian-Lan Zhou zhoujl1999163com

Received 20 August 2013 Accepted 1 October 2013

Academic Editor Zhiguang Feng

Copyright copy 2013 Jian-Lan Zhou et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

Safety risk analysis and assessment of high-risk work system in hydroelectric project has an important role in safety managementThe interactive relationships between human factors and the importance of factors are analyzed and proposed We analyze thecorrelation relationship among the factors by using statistical method which is more objective than subjective judgment TheHFACS is provided to establish a rational and an applicable index system for investigating human error in accidents the structuralequation modeling (SEM) and accident data are used to construct system model and acquire the path coefficient among the riskfactor variables the ANP model is built to assess the importance of accident factors 289 pieces of valid questionnaires data areanalyzed to obtain the path coefficient between risk factor variables and to build the ANP modelrsquos judgment matrix Finally thehuman factorsrsquo weights are calculated by ANP model Combining SEMrsquos results and factorrsquos frequency analysis and building theANPmodel the results show that the four greatest weight values of the factors are respectively ldquopersonal readinessrdquo ldquoperception anddecision errorsrdquo ldquoskill-based errorsrdquo and ldquoviolation operationsrdquo The results of ANP model provide a reference for the engineeringand construction management

1 Introductions

Hydroelectric project construction has higher safety risk forthe interactive factors like complex geological conditionssmall venue large amount of construction workers variousstages of cross-operation and frequent aerial work which isvulnerable to induce safety accidents In recent years in orderto guarantee the safety of hydroelectric project constructionwork a great deal of human material and financial resourceshave been invested the management and supervision of con-struction have gradually strengthened but the overall statusremains grim and the annual total number of hydroelectricproject construction accidents and the number of deathsand serious injuries are still high in China The safety workmanagement is still in a blind state especially the role ofhuman factors in the accident control still lacks clarity Inorder to create a good work environment and improve the

safety index of the hydroelectric project construction weneed to find out the safety factors in construction implementidentification of human risk factors analyze interrelationshipbetween human factors calculate the weight of every factorsfind out the factors need to improve andminimize or preventthe occurrence of accidents

In hydroelectric project construction process a signifi-cant proportion of the technical causes of serious accidentsare attributed to human factors to establish a rational andcomprehensive safety classification system based on humanbehavior is important for safety assessmentThe classificationsystem is used to provide the types of failure involved inaccidents One of the more widely used approaches is theHuman Factors Analysis and Classification System [1] drawnin [2] HFACS is a commonly utilized tool for investi-gating human contributions to aviation accidents under a

2 Mathematical Problems in Engineering

widespread evaluation scheme HFACS and its derivativeshas been adapted applied and promoted in several domains(and countries) in addition to commercial and private flyingincluding mining accident [3 4] helicopter maintenance [5]maritime accident [6 7] railroad accident [8] and surgeryaccident [9] In [6] the HFACSwas extended on an analyticalbasis in a fuzzy environment to investigate shipping accidentsin a consistent manner A sample of 263 significant miningincidents in Australia across 2007-2008 are analyzed usingHFACS and provide a greater understanding of the systemicfactors involved inmining accidents [3]Therefore we extendthe HFACS on an analytical basis in the safety assessment ofwork system in hydropower project construction to evaluatethe faulty behavioral risk value

In HFACS framework extended for hydroelectric projectconstruction there are some observed factors and latentfactors some of these are influenced by each otherquantitative analysis on these factors to assess their weight inwhole system is needed Structural equationmodeling (SEM)is a modeling technique that can handle a large number ofendogenous and exogenous variables as well as latent (unob-served) variables specified as linear combinations (weightedaverages) of the observed variables Regression simultaneousequations (with and without error-term correlations) pathanalysis and variations of factor analysis and canonicalcorrelation analysis are all special cases of SEM [10] Wecan consider the risk as a quantity which can be measuredand expressed by a mathematical relation under the helpof real accidentsrsquo data [11ndash13] SEM is a relatively newmethod and its history can be traced back to the 1970sMost applications have been in psychology sociology thebiological sciences educational research political scienceandmarket research Applications in travel behavior researchdate from 1980 Use of SEM is now rapidly expanding as user-friendly software becomes available and researchers becomecomfortable with SEM and regard it as another tool in theirarsenal Chen et al [14] research the influencing factors ofcoalmine employeesrsquo deliberate violation behaviors in Chinacoalmine fatal accidents

There are some evaluation methods for hydroelec-tric project high-risk operations such as LEC assessmentmethod Safety Inspection Table Analytic Hierarchy Process(AHP) Fault Tree Analysis method Fuzzy ComprehensiveEvaluation method and Neural Network Many scholarsover the world have researched in this area In [15] thedegree of danger was studied when the workers workin potentially dangerous environment presented the LECmethodrsquos formula D = LlowastElowastC where D is the value-at-risk L is the probability of the accident happening E ishow often exposure to dangerous environment and C isthe possible consequences of the accident LEC method isgreatly dependent on the subjectivity of experts which isprone to difference in the process of rating value the resultsare not very objective Dongzhi [16] used Accident TreeAnalysis studied risk factors of hydroelectric engineeringconstruction put forward improvement measures to reducethe incidence of accidents and improved the safety level

of construction But Accident Tree Analysis method hasmany calculation steps and is difficult to make quantitativelyanalysis when the data are less Dedobbeleer and Beland[17] identified the current safety performance evaluationindex of construction work system understood the practicalcharacteristics of workplace by questionnaire survey andaccordingly analyzed construction of safety environmentIn [18 19] safety warnings were proposed after certainsteps including identification of factors which can influencesafety level assessment of potential changes of those factorsassessment of the impact of those changes and selection ofsafety-related criteria

The above studies adopt different evaluation methods toanalyze the project safety but there is no evaluation from aholistic perspective all the methods have some deficienciesApplication of Analytic Network Process (ANP) in theproject construction for safety assessment is a hotspot thisis a method based on Analytic Hierarchical Process (AHP)ANP method considers interrelationship among all factorsin the same level and adjacent levels uses supermatrix tocomprehensively analyze the factors affecting each other andobtains the ultimate hybrid weight In dealing with complexproblems that elements connected with and influenced eachother ANP method is proved to be effective and reasonableby the global studies In [20] fuzzy ANP method wasadopted to evaluate the operation systemrsquos risk factors but thecorrelations among the factors are simply used by the expertsrsquoestimation which may induce expertrsquos bias In [21] the ldquo3P +Irdquo model was proposed to evaluate the effectiveness of safetymanagement system AHP and factor analysis were used toidentify the key indicators impacting the construction andeventually the questionnaire and expert scoringmethod wereadopted to determine the weight In [22] the hydroelectricproject risk factors were studied to establish the index systembased on the ANP and five main classes of risk factors wereidentified organization andmanagement of risks technolog-ical risks natural risks social risk and economic risk andactually a hydroelectric project was assessed In contrast withthe above studies there is a little research on hydroelectricproject construction or it only uses a single method toqualitatively analyze correlation coefficient and may causesubjective influence In [23] it was noted noted that the ANPmethod has some limitations cannot exclude the bias of theexperts the modelrsquos output depends on the given value ofexpert and cause inconsistencies in the pairwise comparisonprocess Therefore it was mentioned that knowledge shouldbe incorporated In [24] it was pointed out should make useof statistical methods for the analysis of accident statisticsso as to more accurately determine dependency relationshipbetween elements which avoid the comparison betweenfactors given by experts with prejudice or inconsistencyproblem

Therefore it is necessary to use ANP method combinedwith quantitative methods and systematically study the fac-tors from the layers of management to construction workersCombining ANP and other methods for comprehensiveassessment can take advantage of their respective advantages

Mathematical Problems in Engineering 3

develop its advantages avoid disadvantages and get betterresults In [25] ANP and Bayesian Networks method wereused to study the safety classification of nuclear powerplants In [26] the ANP and DEMATEL were combinedsuccessfully to solve the evaluation for vehicle fleet main-tenance management In [27] QFD fuzzy ANP and fuzzyFMEA (failure modes and effect analysis) were used toidentify the important types and causes of hazards in theconstruction industry meantime providing risk assessmentvalues of hazard causes and relevant improvement strategiesAbove researches combined ANP with other methods theevaluation process is becoming more refined and morerealistic In this paper we combined ANP HFACS SEM andsynthetic statistical methods to evaluate the high-risk worksystem in hydroelectric projects

The rest of this paper is organized as follows In Section 2the framework of research methodology is constructed andhas been presented in detail In Section 3 based on theHFACS framework the questionnaire is designed and SEMis built by AMOS Section 4 analyzes the correlation factorsrsquointerdependence relationships based on accident cases bylambda method and tau-y method In Section 5 the relativeweights of factors are calculated by synthetic matrix in ANPmodel Finally the results were thoroughly analyzed whilein the last section the main conclusions and future researchtopics were drawn up

2 Methodology Research

ANP model is based on risk influential factorsrsquo classificationand layered architecture This study firstly analyzes thehuman risk factors therefore human factors analysis andclassification system framework (HFACS) is used to analyzehuman factors in construction engineering accidents Thetechnical thinking of this study is firstly applying HFACSand other standardized documents or results to design ques-tionnaire which is designed for the Three Gorges projectand Xiluodu project Xiangjiaba project and then sends thequestionnaire to the management units design units con-struction units supervision units and technical and safetymanagement staff Secondly we analyze the questionnairedata SPSS170 can analyze reliability and validity of the dataand confirm the internal consistency of the data If thedatarsquos reliability is high use AMOS to establish structuralequation modeling (SEM) the path coefficients among thefactors can be obtained thus the relationship can be analyzedamong the factors Thirdly under the HFACS structurethe previous accident cases of Xiluodu project Xiangjiabaproject and theThree Gorges project using statistical meth-ods to analyze human factors of accident we can get thecorrelation coefficient between the factors Finally based onthe preceding analysis combine judgment matrix achievedby empowerment table with judgment matrix by SEM uselinear weighting method obtain one synthesized judgmentmatrix and then calculate this judgment matrix by SuperDecision (SD) tool Eventually we obtain the ANP evaluationweight and ranking of various factors In summary this studywas carried out through interviews questionnaires theoreticanalysis with modeling and statistic methods and decision

and assessment method It consists of 3 stages shown inFigure 1

3 Factors Correlation Analysis Based onEmpirical Study

31 HFACS Framework Before designing the questionnairefirstly make sure of the composition of hydroelectric con-struction risk factors determine the classification and hier-archical structure of human factors and construct hierarchymodel of hydroelectric construction risk then base on themodel to implement the study In this study the HFACSframework is adopted to analyze the human factors whichresult in the engineering construction accidents HFACSconsiders both unsafe behaviors and potential factors whichinfluence unsafe behaviors satisfy the characteristics ofreliability diagnostic and comprehensive in accidents inves-tigation We revise the standard framework of HFACS toadapt with actual safety management of hydroelectric projectconstruction technical measurements personnel quality sit-uation and so forth the adjusted risk influential humanfactor is shown in Figure 2

32 Questionnaire Design In the HFACS framework shownin Figure 1 there are 4 categories and 17 indicators of humanfactors in this study We finally formed a questionnaire with63 detailed items which include 9 items about organizationalinfluences 24 items about safetymanagement 23 items aboutsite work related factors and 7 items about constructionpersonal unsafe behaviors According to the degree of impor-tance the questionnairersquos indicators are in descending orderand adopt Likert-3 table scale method to divide indictors intothree degrees ldquothe first class indicatorrdquo scheduled for score 3ldquothe second class indicatorrdquo scheduled for score 2 and ldquothethird class indicatorrdquo scheduled for score 1 Each item needsto record the corresponding rating value The questionnaireswere issued in 418 pieces 403 valid pieces were collectedAfter sorting and filtering data we finally obtained 289 piecesof valid questionnaires data and based on this tomake validityanalysis

33 Reliability Analysis andValidity Analysis SPSS170 is usedto analyze the reliability and validity of the data By theSPSS softwarersquos ldquoreliability analysisrdquo function the reliabilityanalysis results of all data can be obtained 120572 value is closer to1 the reliability is better Use SPSS softwarersquos ldquofactor analysisrdquofunction to precede validity analysis and get validity resultof all the data the reliability and validity of latent variablesrsquoanalysis results are shown in Table 1 the reliability analysisresults are shown in Table 2 and KMO and Bartlettrsquos valuesare shown in Table 3

In these tables Cronbachrsquos alpha coefficient is the internalconsistency coefficient which is one of the most commonlyused indicators to test questionnairersquos reliability reflecting theconsistency and stability degree of the scale items Bartlettrsquostest assumes that variable correlation coefficient matrix isthe identity matrix if the original hypothesis denied it issuitable for factor analysis KMO is the sampling appropriate


Step

1ststep

Research technical route Research methodInvestigation and

questionnaireEnsure the quality of

questionnaire by FMEA

Extract risk factors calculate factorsrsquo standard value and

standard deviation specify a hierarchical and categorized

structurebuild an ANP safety assessment model

Analyze interactions among the factors2nd

step

Structure accident casestest and modify relationships

among the factors

Build supermatrix weighted supermatrix limitation

supermatrix perform safety

Assess the validity of the modelpropose suggestions based on the assessment monitor key points

revealed by the assessment

Principle component analysis factor analysis

cluster analysis

SPSS software

Analysis by SEM

SPSS and LISREL software

Unstructured data transformation test PRE

method

ANP Method

Super Decision software

Improve weak points revealed by assessmentmonitor key

points revealed by the

assessmentimprove work

systemre-assess

3rd step

SPSS software

assessment

Figure 1 The framework of research methodology

Organizationalinfluences

Organization structure and responsibility Safety investment Safety laws and regulations

L4 layer

Safetymanagement

Education and training

Safety supervision inspection and acceptance

Risk monitoring

Emergency rescue

Accident report investigation and treatment

L2 layer

L3 layer

L1 layer

Site work related factors

Team management

Personal readiness

Mechanical equipment Material

Personal unsafe behaviors

Perception and decision errors Skill-based errorsViolation operations

Technical measurements

Operation environment

Figure 2 Human factors analysis and classification system framework


Table 1 The test result of latent variablesrsquo reliability and validity

Latent variable Measurable variables number KMO Bartlettrsquos test Cronbachrsquos alpha (120572 value)Approx chi-square df Sig

Organizational influences 4 0675 177907 6 0000 0510Safety management 5 0854 536007 10 0000 0822Site work related factors 7 0918 1074889 21 0000 0883Workersrsquo unsafe behaviors 3 0727 342280 3 0000 0816

Table 2 The reliability analysis results of all data

Cronbachrsquos alpha (120572 value) Terms number0916 19

Table 3 The KMO and Bartlettrsquos test results of this study

KMO and Bartlettrsquos testKaiser-Meyer-Olkin measurement of samplingadequacy 0923

Bartlettrsquos testApprox chi-square 2924223df 171Sig 0000

parameter this when the value is greater than 05 means thatthese variables can make factor analysis Sig is significancelevel and less than 005

We can infer from the parameters in the table data that thevalue of 120572 for each subscale is good and the entire question-nairersquos Cronbachrsquos alpha coefficient reaches 0910 close to 1which indicates the high reliability of the questionnaire dataEach subscalersquos KMO and Bartlettrsquos test value is good and theentire questionnaire datarsquos KMO value is 0928 very close to1 Sig lt005 which shows good questionnaire constructionvalidity In short the reliability and validity of the survey dataare desirable

34The Factors Correlation Analysis Based on the SEMModelConsider organizational influences as SEM modelrsquos externallatent variable the corresponding observable variables areexogenous observable variables safety management and thesite work related factors and construction personal unsafebehaviors are latent variable and the corresponding observ-able variables are endogenous observable variable We tryto establish two test models the first model is the highlayer factors which only directly affect their low layers L4effects on L3 L3 effects on L2 L2 effect on L1 (more accordwith the HFACS theory) the second model is L3 affect L2and L1 but L2 does not affect L1 By AMOS170 softwaremake comparison of the two modelsrsquo fit indices the fittingparameter of the first model is more satisfactory and the firstmodel is also more in line with the actual significance of thisstudy Therefore amend the first model and make the resultanalysis

Observe the MI value in the AMOSrsquos output The MIvalue is the revised index which can discover meaningful

information for improving the modelrsquos fitting situation thecorrection index can predict the reduction of the chi-squarevalue Before the correction we must check whether the pathis correct in the model and the variable is really relevantif the regression coefficient is significantly not equal to 0 itrepresents that the path relationship between the variables iscorrect Whenmodifying the model the higher modificationindexrsquos value of the path means more conduciveness toimprove the modelrsquos fitting situation

After repeatedly estimating the model and constantlychecking the output of AMOS software to find out variableswith high MI value simultaneously combine with thepractical significance of the model to increase the correlationpath Eventually we get the fixed model as shown in Figure 3where the path coefficients are marked

Model-fitted indices after being amended are shown inTable 4 We can see that the correction modelrsquos chi-squarevalue is reduced the path value P is significantly below level001 and all fit indices have been improved greatly explainingthe modelrsquos fitting situation that getting better

The correlation coefficient between the variables is over0 which means the relationship between each latent variableis positively correlated indicating that one of the latentvariables will have a positive impact on the other latentvariable Similarly the influence between the latent variableand its corresponding observable variables is positive

We may acquire analysis result by the AMOS thatin the organizational influences layer the safety laws andregulationsrsquo standardization path coefficient is the highest(0799) which indicates the safety laws and regulations havea very big influence in this level In safety managementlayer emergency rescuersquos standardization path coefficient isthe highest (0765) followed by risk monitoring (0755) thenext is education and training (0735) In site work relatedfactors team managementrsquos standardized path coefficient is0802 showing the biggest influence in this layer followed bytechnicalmeasurements (0760) Inworkersrsquo unsafe behaviorslayer perception and decision errorsrsquo path coefficient is thehighest (0901) therefore its influence is themost in this layerfollowed by skill-based errors and violation operations Theinterrelationship between hidden variables is different thecorrelation coefficient between organizational influences andthe safety management the correlation coefficient betweensafety management and site work related factors the correla-tion coefficient between site related factors and constructionpersonal unsafe behaviors are respectively 0872 0808and 0547 therefore the organizational influences have thegreatest impact on safety management


Safety investment



Safety laws and regulations

Risk monitoring

Emergency rescue

Personal readiness

Operating environment

The accident report investigationand treatment

Material

Mechanical equipment

Team management


Perception and decision errors

Skill-based errors

Organization structure and responsibility

Violation operations


Personalunsafe

behaviors


Safety management

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

0872

0808

0547

037004760799

0734072907560755

0639

0631076008020753

0710

0759

09010688

0707

e1

e2

e3

e4

e5

e6

e7

e8

e9

e10

e11

e12

e13

e14

e15

e16

e17

z1

z2

z3

Figure 3 Path analysis graph for SEM revised model

Table 4 Commonly used fitting index computed result of revised model

Fit index Chi-square FID CFI NFI IFI RFI RMSEA AIC BCC GFI RMRResult 181207 108 0970 0930 0970 0911 0049 271207 277207 0932 0124

4 The Correlation Analysis of FactorsBased on Accident Cases

Based on the accident data we count accidents caused byhuman factors find out factor categories with big proportionand analyze their influence on accidentsThe data come fromldquothe Xiluodu project accident cases analysisrdquo ldquothe Xiangjiabaproject accident cases analysisrdquo and ldquotheThreeGorges projectaccident cases analysisrdquo Apply Kappa coefficient analysismethod to analyze 108 accident cases happened in the abovethree projects Determining the human factorsrsquo correspond-ing accident cases and calculating the percentage accountedfor the total number of all accidents this study gets a generalunderstanding of the frequency of occurrence of each factoras well as the weighting among all the factors The weightsof human factors in Table 5 are calculated on the basis offrequency statistics of all factors resulting in the accident

Empowering values in Table 5 will provide an importantreference to build judgment matrix

Subsequently statistically analyze the interaction betweenhuman factors and use Chi-square test to analyze the cor-relation and identify the linkages between factors applyLambda method and Tau-y method to calculate the pro-portional reduction in error (PRE) which is correlationanalysis Both Lambda method and Tau-y method aredirectional statistics and they can determine the degree ofcorrelation between the human factors By these methodswe find out how the factors influence each other and howto form a clue between different levels The more detailedcorrelation analysis based on accident cases can be referredto in our previous work in the reference Here we take anexample as follows the impact of ldquoorganization structure andresponsibilityrdquo on ldquoeducation and trainingrdquo is calculated inTable 6 When the Tau-y value exceeds 010 the correlation


Table 5 The empowerment for each human factor

First class index Weightai Second class index Weight

biNormalizedweight Wi

Organizationalinfluences L4 01254

Organization structure and responsibility 06019 00741Safety investment 03241 00399

Safety laws and regulations 00926 00114

Safetymanagement L3 03123

Education and training 08981 01106Safety supervision inspection and acceptance 0787 00969

Risk monitoring 07222 00889Emergency rescue 00741 00091

Accident report investigation and treatment 00556 00068

Site work relatedfactors L2 03945

Operating environment 06481 00798Technical measurements 0787 00969

Team management 06667 00821Personal readiness 09167 01129

Mechanical equipment 01296 00160Material 00556 00068

Constructionpersonal unsafebehaviors L1

01671Perception and decision errors 05278 00645

Skill-based errors 03426 00422Violation operations 04907 00604

Table 6 The cross table of ldquoorganization structure and responsibil-ityrdquo on ldquoeducation and trainingrdquo

Countq1 ldquoorganization structure andresponsibilityrdquo sumNot resultingin accident (0)

Resulting inaccident (1)

A1 ldquoeducation andtrainingrdquo

Not resulting inaccident (0) 10 1 11

Resulting in accident (1) 33 64 97Sum 43 65 108

relationship is practical when it exceeds 03 the correlationrelationship is strong

1198641 =

[(108 minus 97) lowast 97 + (108 minus 11) lowast 11]

108

= 19759

1198642 =


43

+


65

= 17318

Tau-119910 = 120591119910 = 1198641 minus 11986421198641

=

19759 minus 17318

19759

= 0124

(1)

Based on the correlation analysis we can draw theHFACSframework shown in Figure 4 which reflects the degree ofcorrelation The thick solid lines indicate strong correlationbetween the two factors (the Tau-y value exceeds 01) and thedashed line indicates the weak correlation between the twofactors In Figure 4 the dashed box means the frequency ofthe occurring factor in the accident cases is less than 01

In Figure 4 there are some connections between the fac-tors ldquoorganization structure and responsibilityrdquo in the L4 layerand ldquoeducation trainingrdquo ldquosafety supervision inspection andacceptancerdquo and ldquoemergency rescuerdquo in the L3 layer the rela-tionship between ldquoorganization structure and responsibilityrdquoand ldquoemergency rescuerdquo is weak which means that safetymanagement facilities safety management personnel andsafe work responsibility system have limited impact on safetywork emergency management and accident rescue but cangreatly affect on the staff ldquoeducation and trainingrdquo and ldquosafetysupervision inspection and acceptancerdquowhich indicates thatsafety managers responsibilitiesrsquo full fulfillments can improvethe effect of safety education and training carefully foundhidden danger strict rectification and process monitoringcan also play an important role in accident prevention

ldquoEducation and trainingrdquo in L3 layer has relationshipwith ldquoteam managementrdquo and ldquopersonal readinessrdquo in L2but the correlation with ldquoteam managementrdquo is weakerwhich indicates that good safety education training of teammembers has a positive effect on good information commu-nication team cooperation and effectiveness of foreknowingdangerous activities The correlation between ldquoeducationtrainingrdquo and ldquopersonal readinessrdquo is strong which meansthat ldquoeducation trainingrdquo can greatly improve the ldquopersonnelrsquosbasic situationrdquo the workers get enough safety education andskills training which enhance their safety consciousness they


Organization structure and responsibility Safety investment

Material




Emergency rescue

Accident report investigation and

treatment

Skill-based errors

Personal basic situation



Team management

Risk monitoring

Safety supervision inspection and

acceptance



0124

0131 0076

00470616

0173 00400189

00550063

00390052

L4 layer

L3 layer

L2 layer

L1 layer

Figure 4 The correlation analysis among the HFACS factors using Tau-ymethod

also can understand their objective situation and avoid beinginvolved in the accidents ldquoSafety supervision inspection andacceptancerdquo in L3 layers and ldquopersonal readinessrdquo in L2 layeras well as ldquomechanical equipmentrdquo have relationships whichmean ldquosafety supervision inspection and acceptancerdquo affectsboth the workersrsquo situation and the mechanical equipmentsafety management but less the latter ldquoRisk controlrdquo inL3 layer and ldquotechnical measuresrdquo in L2 layers also haverelationship which means that the dangerous places andhazards identification assessment and monitoring can leadto more targeted and practical measures The premise ofthe safety warning signs set is the hazards identificationthe rational allocation of safety measurements and confidingtechnical intentions are also determined by the hazardsidentification

There are relationships between ldquooperation environmentrdquoin layer 2 and ldquoperception and decision errorsrdquo in layer 1ldquotechnical measuresrdquo in layer 2 and ldquoskill-based errorsrdquo inlayer 1 ldquoPersonal readinessrdquo in layer 2 and ldquoperception anddecision errorsrdquo ldquoviolation operationrdquo in layer 1 The dottedlines mean the relationships are weak indicating that theconstruction workersrsquo unsafe behavior is little affected bysite work related conditions The capacity of the workerrsquosperception and decision-making work skills and operationalviolations are affected by the individual subjective individualtechnical ability and accidental factors therefore there aresome relationships between L2 layer factors and L1 layer ones

5 Safety Assessment Based onthe ANP Method

51 Molding and Building Judgment Matrix According tothe HFACS framework as well as the mutual correlationamong the human factors build the ANP network hierarchyevaluation model as shown in Figure 5 The model reflectsthe relationship between the various factors in the criterionlayer

The core work of the ANPrsquos empowerment and solutionis to compute each supermatrix weighted super matrix and

limitation supermatrix which is a very complex calculationprocess Therefore we use the Super Decision tool to dealwith the calculation

The judgment matrix constructed in this study is quitedifferent from other studies The judgment matrix is notfrom the expertrsquos pairwise comparison but linearly weighs thejudgmentmatrix1198821015840 and judgmentmatrix11988210158401015840Thenext bothmatrixes are respectively from the pairwise comparison ofempowerment values (see Table 5) and the pairwise compar-ison of path coefficients of structure equation modeling (seeFigure 3) According to the properties of the positive recip-rocal matrix use the following formula to obtain syntheticmatrix

119882 = 1205721198821015840+ (1 minus 120572)119882

10158401015840 (2)

In this formula 120572 is weighted index 120572 isin [0 1] 1198821015840isbuilt by the pairwise comparison of empowerment valuesin Table 5 11988210158401015840 is built by the pairwise comparison of pathcoefficients of structure equation modeling in Figure 3 and119882 is the final judgment matrix 119882101584011988210158401015840 and 119882 are allpositive reciprocal matrixes subjected to 119886

119894119895gt 0119886

119894119894=

1119886119894119895= 1119886

119895119894(119894 119895 = 1 2 119899) The judgment matrix is from

concrete values compared with each other so the judgmentmatrix is satisfied with 119886

119894119895= 119886119894119896119886119895119896 Each judgment matrixrsquo

consistency ratio CR is equal to zero and is satisfied with fullconsistency Using the synthetic matrix the ANP assessmentprocess is a fully quantitative process

The value of weighted index 120572 is set to 07 on preferenceAll factors of layers with mutual relationship are carried outpair-wise comparisons The detailed calculation process is asfollows

Firstly build the judgment matrixes of ldquoorganizationalinfluencesrdquo ldquosafety managementrdquo ldquosite work related factorsrdquoand ldquopersonal unsafe behaviorsrdquo

1198821015840

1=

[

[

[

[

1 040 032 075

25 1 079 186

3125 1266 1 235

133 0538 0426 1

]

]

]

]


Safety evaluation of high-risk operations

Organizational influences

Safetymanagement


Construction personal unsafe behaviors

Org

aniz

atio

n str

uctu

re an

d re

spon

sibili

ty

Educ

atio

n an

d tr

aini

ng

Safe

ty la

ws a

nd re

gulat

ions

Safe

ty in

vestm

ent

Risk

mon

itorin

g

Pers

onal

bas

ic si

tuat

ion

Perc

eptio

n an

d de

cisio

n er

rors

Skill

-bas

ed er

rors

Team

man

agem

ent

Safe

ty su

perv

ision

ins

pect

ion

and

acce

ptan

ce

Mat

eria

l

Tech

nica

l mea

sure

men

ts

Acci

dent

repo

rt i

nves

tigat

ion

and

treat

men

t

Ope

ratin

g en

viro

nmen

t

Emer

genc

y re

scue

Viol

atio

n op

erat

ions

Mec

hani

cal e

quip

men

tFigure 5 Hierarchical and correlation of the factors in ANP model

11988210158401015840

1=

[

[

[

[

1 1147 1418 2597

0872 1 1238 2262

0705 0808 1 1828

0385 0442 0547 1

]

]

]

]

(3)

According to the formula (2) the synthetic matrix is asfollows

1198821=

[

[

[

[

1 062 065 130

161 1 092 198

154 109 1 219

077 051 046 1

]

]

]

]

(4)

The judgment matrix of the elements in ldquoorganizationalinfluencesrdquo is as follows

1198822=[

[

1 154 469

065 1 263

021 038 1

]

]

(5)

The judgment matrix of the elements in ldquosafety manage-mentrdquo is as follows

1198823=

[

[

[

[

[

[

1 110 116 880 1173

091 1 105 775 1032

086 095 1 714 950

011 013 014 1 129

009 010 011 078 1

]

]

]

]

]

]

(6)

The judgment matrix of the elements in ldquosite work relatedfactorsrdquo is as follows

1198824=

[

[

[

[

[

[

[

[

1 082 092 075 376 847

122 1 111 091 456 1028

109 090 1 083 393 877

133 110 120 1 526 1192

027 022 025 019 1 193

012 010 011 008 052 1

]

]

]

]

]

]

]

]

(7)

The judgment matrix of the elements in ldquopersonal unsafebehaviorsrdquo is as follows

1198825=[

[

1 147 114

068 1 078

088 128 1

]

]

(8)

Considering correlation analysis the judgment matrix ofldquoorganizational structure and responsibilitiesrdquo to its correla-tion factors is as follows

1198826=[

[

1 0947 1632

1056 1 1724

0613 0580 1

]

]

(9)

Considering correlation analysis the judgment matrix ofldquoeducation and trainingrdquo to its correlation factors is as follows

1198827= [

1 0076

13158 1

] (10)


Table 7 ANP assessment weights

First class index ANP weight Second class index ANPweight ANP rank Normalized

weight Wi Cases rank

Organizationalinfluences 01315

q1 organization structure and responsibility 00347 11 00741 8q2 safety investment 00214 13 00399 12

q3 safety laws and regulations 00078 15 00114 14

SafetyManagement 03474

a1 education and training 00693 6 01106 2a2 safety supervision inspection and acceptance 00652 7 00969 3

a3 risk monitoring 00493 8 00889 5a4 emergency rescue 00146 14 00091 15

a5 accident report investigation and treatment 00050 16 00068 16

Site work relatedfactors 03755

x1 operating environment 00363 10 00798 7x2 technical measurements 00934 5 00969 4

x3 team management 00440 9 00821 6x4 personal readiness 01664 1 01129 1

x5 mechanical equipment 00216 12 00160 13x6 material 00044 17 00068 17

Constructionpersonal unsafebehaviors

01456d1 perception and decision errors 01551 2 0065 9

d2 skill-based errors 01122 3 00422 11d3 violation operations 00994 4 00604 10

Considering correlation analysis the judgment matrixof ldquosafety supervision inspection and acceptancerdquo to itscorrelation factors is as follows

1198828= [

1 4325

0231 1

] (11)

Considering correlation analysis the judgment matrixof ldquopersonal basic situationrdquo to its correlation factors is asfollows

1198829= [

1 1212

0825 1

] (12)

According to Figure 5 we use SD tool to build theANP model The model reflects the relationship betweenthe variables in the layer factors At the network layer wehave four categories each category has several elements (17evaluation indicators in the sum) Because the factors in thelayers are not independent the circular arrow lines are seenin Figure 5

52 Solutions Through calculation by the SD software theweight values of every factor are shown in Table 7

53 Results In Table 7 the four smallest weight values ofthe factors are respectively ldquomaterialrdquo (00044) ldquoaccidentreport investigation and treatmentrdquo (00050) ldquosafety lawsand regulationsrdquo (00078) and ldquoemergency rescuerdquo (00146)The normalized weight values based on cases statistics alsoshow that these four factors result in accidents less frequentlywhich indicate that these four factors less likely to result inaccidents in the high-risk construction operations and theorganizations have done well in these four aspects

The four greatest weight values of the factors arerespectively ldquopersonal readinessrdquo (01664) ldquoperception anddecision errorsrdquo (01551) ldquoskill-based errorsrdquo (01122) andldquoviolation operationsrdquo (00994) But in of cases statisti-cal analysis the four greatest weight values are ldquopersonalreadinessrdquo(01129) ldquoeducation trainingrdquo (01106) ldquotechnicalmeasuresrdquo (00969) and ldquosafety supervision inspection andacceptancerdquo (00969) Only ldquopersonal readinessrdquo is the mostgreatest in both methods which shows that in the projectconstruction when the workerrsquos basic situation greatly influ-ences his safety consciousness risk awareness and psycho-logical andphysiological conditions In order to guarantee thesafety of construction projects organizations should strive toimprove this factor The rank of ldquoeducating trainingrdquo dropsfrom the original 2 to 6 indicating that the interaction amongthe factors will lead to the assessment results change Becausethe ldquoeducation and trainingrdquo and ldquopersonal readinessrdquo havea very strong relationship the imperfections of the safetyeducation and skills training will lead to personnelrsquos basicsituation get worse In order to avoid the double counting ofthe associated factors the assessment weight of ldquoeducationand trainingrdquo decreases The weight of ldquosafety supervisioninspection and acceptancerdquo drops from the original ranking3 to 7 which is a result that this factor also directly affectsldquopersonal readinessrdquo So with the similar reason the ANPassessment weight of ldquosafety supervision inspection andacceptancerdquo decreases

After ANP assessment the weight values of ldquoperceptionand decision errorsrdquo ldquoskill-based errorsrdquo and ldquooperationviolationrdquo have increased According to the results of fac-tor analysis ldquoeducation trainingrdquo and ldquosafety supervisioninspection and acceptancerdquo will influence ldquopersonal readi-nessrdquo ldquopersonal readinessrdquo located in L2 layer directly influ-ences ldquoperception and decision-making errorsrdquo in L1 layer


and ldquooperation violationrdquo and ldquorisk monitoringrdquo influencesldquotechnology measuresrdquo ldquotechnology measuresrdquo in L2 layerinfluence ldquoskills errorsrdquo in L1 layer It can be seen thatconstruction workerrsquos unsafe behavior is the direct influentialfactor which may lead to the accident Three factors in L1layer ldquoperception and decision-making errorsrdquo ldquoskill errorsrdquoand ldquooperation irregularitiesrdquo with higher weight values inANP assessment which also indicate the unsafe behavior ofconstruction workers are the most important factors leadingto accidents

In ldquoorganizational influencesrdquo layer the three factorsof ldquoorganization structure and dutiesrdquo ldquosafety investmentsrdquoand ldquosafety laws and regulationsrdquo their weight values beingrelatively smaller indicate that the organizations have takencomplete measurements on these aspects have invested onsafety management institutions safety management per-sonnel and have established the safety work responsibilitysystem series of laws and regulations and relevant rules andregulations All the above measures are successful

In the actual construction project these 17 assessmentfactors often influence each other so the ANP assessmentresults may be more realistic and can provide a reference forthe engineering and construction management Meanwhilethere are still a lot of factors need to be considered todetermine the final management plans and schedules

6 Conclusions

This study firstly revises the standard HFACS frameworkto evaluate the risk factors of the high-risk operations inhydroelectric engineering construction constructs a compre-hensive framework system from the organizational layer topersonal layer and is based on the framework to deal withthe subsequent research

Secondly this study obtains the original data from ques-tionnaire and analyzes the data by the SPSS The reliabilityand validity analysis results indicate that the questionnairedata met the realistic requirements The conceptual modelis drawn by AMOS the raw data is imported from theSPSS to fit make comparison revise and analyze the modelAfter modeling analyzing and revising we get correlationcoefficients between latent variables which may influencehydroelectric construction safety as well as correlation coef-ficients between latent variables and their correspondingobservable variables The correlation coefficients excess zerowhich means the variables have positive relationships if anyvariable (factor) is improved other variables (factors) willalso be improved to some degree The value of correlationcoefficient between variables shows the influence on eachother These results give some reference for the organizationsto develop management regulations and strategies

Thirdly we use the statistical methods such as the PREmethod revise HFACS framework to analyze 108 accidentcases and count the frequency of each risk factor in theaccidentsWe use the chi-square test to determine correlationbetween adjacent level factors in order to determine theconcrete association degree between the factors more accu-rately and calculate the correlation coefficient with the PRE

method between the factors The coefficient values indicatethe correlation degree between the two factors

Finally we use the ANP method to evaluate the impor-tance of the factors influencing safety work The traditionalsafety assessment methods generally use subjective qualita-tive or semiqualitative principles not quantitatively assessthe safety and risk of construction project The AHP methodcannot consider the interrelationship between the factorsthey do and is not consistent with the actual situationHowever the ANP method makes up for such deficiencyIn this study the ANP modelrsquos judgment matrix is not fromthe pair-wise comparison method but from a combinationof accident cases analysis results of factor frequency thecorrelation coefficient between the factors and the pathcoefficient of structural equation modeling Then we followa linear formula to get the final judgment matrix whichcan improve the qualitative analysis result relative to thetraditional ANP method (the expert rating) Such methodmakes possible the assessment results more objective andquantitative

Due to research limitation there remains a furtheranalysis to satisfy a more realistic factors classification andhierarchical relationships as well asmore rational frameworkThe accident cases data are also limited and cannot cover allcharacteristics of risk factors The analysis model is to someextent simple according to a fixed direction to make factoranalysis and the variables in the analysis process are nominalvariables However in the actual construction project therelationships between the factors are complex there are nosuch simple relationships in the HFACS model Thereforethere may some deviations between the analysis results andthe realistic situation

Finally this study only selects structural equation model-ing accidents statistical analysis and ANP method to imple-ment the safety assessment research and has not compre-hensively compared other more methods such as Bayesiantheory D-S evidence theory and neural network Thereforethe assessment results may not be most accurate and optimalAs a result we should carry out a variety of assessmentmethods and select the combination of optimal methods toevaluate in the future

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

This paper is supported by the National Natural ScienceFund Project (50909045 51079078) and the Fundamen-tal Research Funds for the Central Universities (HUST2013QN154)

References

[1] D A Wiegmann and S A Shappell ldquoHuman error analysis ofcommercial aviation accidents application of the human factors


analysis and classification system (HFACS)rdquoAviation Space andEnvironmental Medicine vol 72 no 11 pp 1006ndash1016 2001

[2] J Reason Human Error Cambridge University Press NewYork NY USA 1990

[3] M G Lenne PM Salmon C C Liu andM Trotter ldquoA systemsapproach to accident causation in mining an application of theHFACS methodrdquo Accident Analysis and Prevention vol 48 pp111ndash117 2012

[4] J M Patterson and S A Shappell ldquoOperator error and systemdeficiencies analysis of 508 mining incidents and accidentsfrom Queensland Australia using HFACSrdquo Accident Analysisand Prevention vol 42 no 4 pp 1379ndash1385 2010

[5] H S J Rashid C S Place and G R Braithwaite ldquoHelicoptermaintenance error analysis beyond the third order of theHFACS-MErdquo International Journal of Industrial Ergonomicsvol 40 no 6 pp 636ndash647 2010

[6] M Celik and S Cebi ldquoAnalytical HFACS for investigatinghuman errors in shipping accidentsrdquo Accident Analysis andPrevention vol 41 no 1 pp 66ndash75 2009

[7] C Chauvin S Lardjane G Morel and J P ClostermannldquoHuman and organizational factors in maritime accidentsanalysis of collisions at sea using the HFACSrdquo Accident Analysisand Prevention vol 59 pp 26ndash37 2013

[8] S Reinach and A Viale ldquoApplication of a human errorframework to conduct train accidentincident investigationsrdquoAccident Analysis and Prevention vol 38 no 2 pp 396ndash4062006

[9] AW ElBardissi D AWiegmann J A Dearani R C Daly andT M Sundt III ldquoApplication of the human factors analysis andclassification systemmethodology to the cardiovascular surgeryoperating roomrdquo Annals of Thoracic Surgery vol 83 no 4 pp1412ndash1419 2007

[10] T F Golob ldquoStructural equation modeling for travel behaviorresearchrdquo Transportation Research B vol 37 no 1 pp 1ndash252003

[11] P K Marhavilas and D Koulouriotis ldquoRisk Estimation in theConstructionsrsquo Worksites by using a Quantitative AssessmentTechnique and Statistical Information of Accidentsrdquo ScientificJournal of Technical Chamber of Greece vol 1 no 1-2 pp 47ndash602007

[12] P K Marhavilas and D E Koulouriotis ldquoA risk-estimationmethodological framework using quantitative assessment tech-niques and real accidentsrsquo data application in an aluminumextrusion industryrdquo Journal of Loss Prevention in the ProcessIndustries vol 21 no 6 pp 596ndash603 2008

[13] P K Marhavilas D E Koulouriotis and K VoulgaridouldquoDevelopment of a quantitative risk assessment technique andapplication on an industryrsquos worksite using real accidentsrsquo datardquoScientific Journal of Hellenic Association of Mechanical andElectrical Engineers vol 416 pp 14ndash20 2009

[14] H Chen H Qi O Wang and R-Y Long ldquoThe research on thestructural equation model of affecting factors of deliberate vio-lation in coalmine fatal accidents in Chinardquo System EngineeringTheory and Practice vol 27 no 8 pp 127ndash136 2007

[15] K J Graham andG F Kinney ldquoExplosive shocks in airrdquo Journalof the Acoustical Society of America vol 80 no 2 pp 708ndash7091986

[16] Peng Dongzhi ldquoFour dangers condition identification andcontrol in water and electricity project construction worksystemrdquo Construction Technique vol 26 no 5 pp 70ndash72 2007

[17] N Dedobbeleer and F Beland ldquoA safety climate measure forconstruction sitesrdquo Journal of Safety Research vol 22 no 2 pp97ndash103 1991

[18] K Oslashien ldquoRisk indicators as a tool for risk controlrdquo ReliabilityEngineering and System Safety vol 74 no 2 pp 129ndash145 2001


[20] M Dagdeviren I Yuksel and M Kurt ldquoA fuzzy analyticnetwork process (ANP) model to identify faulty behavior risk(FBR) in work systemrdquo Safety Science vol 46 no 5 pp 771ndash783 2008

[21] E Ai Lin Teo and F Yean Yng Ling ldquoDeveloping a modelto measure the effectiveness of safety management systems ofconstruction sitesrdquo Building and Environment vol 41 no 11 pp1584ndash1592 2006

[22] D Zhong S Cai and Y Li ldquoRisk analysis of hydropower projectbased on analytic network process and its applicationrdquo Journalof Hydroelectric Engineering vol 27 no 1 pp 11ndash17 2008

[23] Z Ayag and R G Ozdemir ldquoA hybrid approach to conceptselection through fuzzy analytic network processrdquo Computersand Industrial Engineering vol 56 no 1 pp 368ndash379 2009

[24] K F R Liu and J-H Lai ldquoDecision-support for environmentalimpact assessment a hybrid approach using fuzzy logic andfuzzy analytic network processrdquo Expert Systems with Applica-tions vol 36 no 3 pp 5119ndash5136 2009

[25] J S Ha and P H Seong ldquoA method for risk-informed safetysignificance categorization using the analytic hierarchy processand bayesian belief networksrdquo Reliability Engineering and Sys-tem Safety vol 83 no 1 pp 1ndash15 2004

[26] D Vujanovic V Momcilovic N Bojovic and V Papic ldquoEval-uation of vehicle fleet maintenance management indicatorsby application of DEMATEL and ANPrdquo Expert Systems withApplications vol 39 no 12 pp 10552ndash10563 2012

[27] H-T Liu and Y-L Tsai ldquoA fuzzy risk assessment approachfor occupational hazards in the construction industryrdquo SafetyScience vol 50 no 4 pp 1067ndash1078 2012

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MathematicsJournal of


Mathematical Problems in Engineering

Hindawi Publishing Corporationhttpwwwhindawicom

Differential EquationsInternational Journal of

Volume 2014

Applied MathematicsJournal of


Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of


Mathematical PhysicsAdvances in

Complex AnalysisJournal of


OptimizationJournal of


CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of


Operations ResearchAdvances in

Journal of


Function Spaces

Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of Mathematics and Mathematical Sciences


The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014


Algebra

Discrete Dynamics in Nature and Society



Decision SciencesAdvances in

Discrete MathematicsJournal of


Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Stochastic AnalysisInternational Journal of


widespread evaluation scheme HFACS and its derivativeshas been adapted applied and promoted in several domains(and countries) in addition to commercial and private flyingincluding mining accident [3 4] helicopter maintenance [5]maritime accident [6 7] railroad accident [8] and surgeryaccident [9] In [6] the HFACSwas extended on an analyticalbasis in a fuzzy environment to investigate shipping accidentsin a consistent manner A sample of 263 significant miningincidents in Australia across 2007-2008 are analyzed usingHFACS and provide a greater understanding of the systemicfactors involved inmining accidents [3]Therefore we extendthe HFACS on an analytical basis in the safety assessment ofwork system in hydropower project construction to evaluatethe faulty behavioral risk value

In HFACS framework extended for hydroelectric projectconstruction there are some observed factors and latentfactors some of these are influenced by each otherquantitative analysis on these factors to assess their weight inwhole system is needed Structural equationmodeling (SEM)is a modeling technique that can handle a large number ofendogenous and exogenous variables as well as latent (unob-served) variables specified as linear combinations (weightedaverages) of the observed variables Regression simultaneousequations (with and without error-term correlations) pathanalysis and variations of factor analysis and canonicalcorrelation analysis are all special cases of SEM [10] Wecan consider the risk as a quantity which can be measuredand expressed by a mathematical relation under the helpof real accidentsrsquo data [11ndash13] SEM is a relatively newmethod and its history can be traced back to the 1970sMost applications have been in psychology sociology thebiological sciences educational research political scienceandmarket research Applications in travel behavior researchdate from 1980 Use of SEM is now rapidly expanding as user-friendly software becomes available and researchers becomecomfortable with SEM and regard it as another tool in theirarsenal Chen et al [14] research the influencing factors ofcoalmine employeesrsquo deliberate violation behaviors in Chinacoalmine fatal accidents

There are some evaluation methods for hydroelec-tric project high-risk operations such as LEC assessmentmethod Safety Inspection Table Analytic Hierarchy Process(AHP) Fault Tree Analysis method Fuzzy ComprehensiveEvaluation method and Neural Network Many scholarsover the world have researched in this area In [15] thedegree of danger was studied when the workers workin potentially dangerous environment presented the LECmethodrsquos formula D = LlowastElowastC where D is the value-at-risk L is the probability of the accident happening E ishow often exposure to dangerous environment and C isthe possible consequences of the accident LEC method isgreatly dependent on the subjectivity of experts which isprone to difference in the process of rating value the resultsare not very objective Dongzhi [16] used Accident TreeAnalysis studied risk factors of hydroelectric engineeringconstruction put forward improvement measures to reducethe incidence of accidents and improved the safety level

of construction But Accident Tree Analysis method hasmany calculation steps and is difficult to make quantitativelyanalysis when the data are less Dedobbeleer and Beland[17] identified the current safety performance evaluationindex of construction work system understood the practicalcharacteristics of workplace by questionnaire survey andaccordingly analyzed construction of safety environmentIn [18 19] safety warnings were proposed after certainsteps including identification of factors which can influencesafety level assessment of potential changes of those factorsassessment of the impact of those changes and selection ofsafety-related criteria

The above studies adopt different evaluation methods toanalyze the project safety but there is no evaluation from aholistic perspective all the methods have some deficienciesApplication of Analytic Network Process (ANP) in theproject construction for safety assessment is a hotspot thisis a method based on Analytic Hierarchical Process (AHP)ANP method considers interrelationship among all factorsin the same level and adjacent levels uses supermatrix tocomprehensively analyze the factors affecting each other andobtains the ultimate hybrid weight In dealing with complexproblems that elements connected with and influenced eachother ANP method is proved to be effective and reasonableby the global studies In [20] fuzzy ANP method wasadopted to evaluate the operation systemrsquos risk factors but thecorrelations among the factors are simply used by the expertsrsquoestimation which may induce expertrsquos bias In [21] the ldquo3P +Irdquo model was proposed to evaluate the effectiveness of safetymanagement system AHP and factor analysis were used toidentify the key indicators impacting the construction andeventually the questionnaire and expert scoringmethod wereadopted to determine the weight In [22] the hydroelectricproject risk factors were studied to establish the index systembased on the ANP and five main classes of risk factors wereidentified organization andmanagement of risks technolog-ical risks natural risks social risk and economic risk andactually a hydroelectric project was assessed In contrast withthe above studies there is a little research on hydroelectricproject construction or it only uses a single method toqualitatively analyze correlation coefficient and may causesubjective influence In [23] it was noted noted that the ANPmethod has some limitations cannot exclude the bias of theexperts the modelrsquos output depends on the given value ofexpert and cause inconsistencies in the pairwise comparisonprocess Therefore it was mentioned that knowledge shouldbe incorporated In [24] it was pointed out should make useof statistical methods for the analysis of accident statisticsso as to more accurately determine dependency relationshipbetween elements which avoid the comparison betweenfactors given by experts with prejudice or inconsistencyproblem

Therefore it is necessary to use ANP method combinedwith quantitative methods and systematically study the fac-tors from the layers of management to construction workersCombining ANP and other methods for comprehensiveassessment can take advantage of their respective advantages













Step

1ststep








step


among the factors






cluster analysis

SPSS software

Analysis by SEM



method

ANP Method





systemre-assess

3rd step

SPSS software

assessment




L4 layer

Safetymanagement



Risk monitoring

Emergency rescue


L2 layer

L3 layer

L1 layer


Team management

Personal readiness


























Safety investment




Risk monitoring

Emergency rescue

Personal readiness



Material


Team management



Skill-based errors




Personalunsafe

behaviors


Safety management

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

0872

0808

0547

037004760799

0734072907560755

0639

0631076008020753

0710

0759

09010688

0707

e1

e2

e3

e4

e5

e6

e7

e8

e9

e10

e11

e12

e13

e14

e15

e16

e17

z1

z2

z3

































1198641 =


108

= 19759

1198642 =


43

+


65

= 17318

Tau-119910 = 120591119910 = 1198641 minus 11986421198641

=

19759 minus 17318

19759

= 0124

(1)






Material




Emergency rescue


treatment

Skill-based errors




Team management

Risk monitoring


acceptance



0124

0131 0076

00470616

0173 00400189

00550063

00390052

L4 layer

L3 layer

L2 layer

L1 layer









119882 = 1205721198821015840+ (1 minus 120572)119882

10158401015840 (2)


119894119895gt 0119886

119894119894=

1119886119894119895= 1119886







1198821015840

1=

[

[

[

[

1 040 032 075

25 1 079 186

3125 1266 1 235

133 0538 0426 1

]

]

]

]




Safetymanagement



Org

aniz

atio

n str

uctu

re an

d re

spon

sibili

ty

Educ

atio

n an

d tr

aini

ng

Safe

ty la

ws a

nd re

gulat

ions

Safe

ty in

vestm

ent

Risk

mon

itorin

g

Pers

onal

bas

ic si

tuat

ion

Perc

eptio

n an

d de

cisio

n er

rors

Skill

-bas

ed er

rors

Team

man

agem

ent

Safe

ty su

perv

ision

ins

pect

ion

and

acce

ptan

ce

Mat

eria

l

Tech

nica

l mea

sure

men

ts

Acci

dent

repo

rt i

nves

tigat

ion

and

treat

men

t

Ope

ratin

g en

viro

nmen

t

Emer

genc

y re

scue

Viol

atio

n op

erat

ions

Mec

hani

cal e

quip

men


11988210158401015840

1=

[

[

[

[

1 1147 1418 2597

0872 1 1238 2262

0705 0808 1 1828

0385 0442 0547 1

]

]

]

]

(3)


1198821=

[

[

[

[

1 062 065 130

161 1 092 198

154 109 1 219

077 051 046 1

]

]

]

]

(4)


1198822=[

[

1 154 469

065 1 263

021 038 1

]

]

(5)


1198823=

[

[

[

[

[

[

1 110 116 880 1173

091 1 105 775 1032

086 095 1 714 950

011 013 014 1 129

009 010 011 078 1

]

]

]

]

]

]

(6)


1198824=

[

[

[

[

[

[

[

[

1 082 092 075 376 847

122 1 111 091 456 1028

109 090 1 083 393 877

133 110 120 1 526 1192

027 022 025 019 1 193

012 010 011 008 052 1

]

]

]

]

]

]

]

]

(7)


1198825=[

[

1 147 114

068 1 078

088 128 1

]

]

(8)


1198826=[

[

1 0947 1632

1056 1 1724

0613 0580 1

]

]

(9)


1198827= [

1 0076

13158 1

] (10)




















1198828= [

1 4325

0231 1

] (11)


1198829= [

1 1212

0825 1

] (12)










6 Conclusions










Acknowledgments


References





































Volume 2014




Journal of











Journal of


Function Spaces






Algebra





















Step

1ststep








step


among the factors






cluster analysis

SPSS software

Analysis by SEM



method

ANP Method





systemre-assess

3rd step

SPSS software

assessment




L4 layer

Safetymanagement



Risk monitoring

Emergency rescue


L2 layer

L3 layer

L1 layer


Team management

Personal readiness


























Safety investment




Risk monitoring

Emergency rescue

Personal readiness



Material


Team management



Skill-based errors




Personalunsafe

behaviors


Safety management

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

0872

0808

0547

037004760799

0734072907560755

0639

0631076008020753

0710

0759

09010688

0707

e1

e2

e3

e4

e5

e6

e7

e8

e9

e10

e11

e12

e13

e14

e15

e16

e17

z1

z2

z3

































1198641 =


108

= 19759

1198642 =


43

+


65

= 17318

Tau-119910 = 120591119910 = 1198641 minus 11986421198641

=

19759 minus 17318

19759

= 0124

(1)






Material




Emergency rescue


treatment

Skill-based errors




Team management

Risk monitoring


acceptance



0124

0131 0076

00470616

0173 00400189

00550063

00390052

L4 layer

L3 layer

L2 layer

L1 layer









119882 = 1205721198821015840+ (1 minus 120572)119882

10158401015840 (2)


119894119895gt 0119886

119894119894=

1119886119894119895= 1119886







1198821015840

1=

[

[

[

[

1 040 032 075

25 1 079 186

3125 1266 1 235

133 0538 0426 1

]

]

]

]




Safetymanagement



Org

aniz

atio

n str

uctu

re an

d re

spon

sibili

ty

Educ

atio

n an

d tr

aini

ng

Safe

ty la

ws a

nd re

gulat

ions

Safe

ty in

vestm

ent

Risk

mon

itorin

g

Pers

onal

bas

ic si

tuat

ion

Perc

eptio

n an

d de

cisio

n er

rors

Skill

-bas

ed er

rors

Team

man

agem

ent

Safe

ty su

perv

ision

ins

pect

ion

and

acce

ptan

ce

Mat

eria

l

Tech

nica

l mea

sure

men

ts

Acci

dent

repo

rt i

nves

tigat

ion

and

treat

men

t

Ope

ratin

g en

viro

nmen

t

Emer

genc

y re

scue

Viol

atio

n op

erat

ions

Mec

hani

cal e

quip

men


11988210158401015840

1=

[

[

[

[

1 1147 1418 2597

0872 1 1238 2262

0705 0808 1 1828

0385 0442 0547 1

]

]

]

]

(3)


1198821=

[

[

[

[

1 062 065 130

161 1 092 198

154 109 1 219

077 051 046 1

]

]

]

]

(4)


1198822=[

[

1 154 469

065 1 263

021 038 1

]

]

(5)


1198823=

[

[

[

[

[

[

1 110 116 880 1173

091 1 105 775 1032

086 095 1 714 950

011 013 014 1 129

009 010 011 078 1

]

]

]

]

]

]

(6)


1198824=

[

[

[

[

[

[

[

[

1 082 092 075 376 847

122 1 111 091 456 1028

109 090 1 083 393 877

133 110 120 1 526 1192

027 022 025 019 1 193

012 010 011 008 052 1

]

]

]

]

]

]

]

]

(7)


1198825=[

[

1 147 114

068 1 078

088 128 1

]

]

(8)


1198826=[

[

1 0947 1632

1056 1 1724

0613 0580 1

]

]

(9)


1198827= [

1 0076

13158 1

] (10)




















1198828= [

1 4325

0231 1

] (11)


1198829= [

1 1212

0825 1

] (12)










6 Conclusions










Acknowledgments


References





































Volume 2014




Journal of











Journal of


Function Spaces






Algebra




























Safety investment




Risk monitoring

Emergency rescue

Personal readiness



Material


Team management



Skill-based errors




Personalunsafe

behaviors


Safety management

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

0872

0808

0547

037004760799

0734072907560755

0639

0631076008020753

0710

0759

09010688

0707

e1

e2

e3

e4

e5

e6

e7

e8

e9

e10

e11

e12

e13

e14

e15

e16

e17

z1

z2

z3

































1198641 =


108

= 19759

1198642 =


43

+


65

= 17318

Tau-119910 = 120591119910 = 1198641 minus 11986421198641

=

19759 minus 17318

19759

= 0124

(1)






Material




Emergency rescue


treatment

Skill-based errors




Team management

Risk monitoring


acceptance



0124

0131 0076

00470616

0173 00400189

00550063

00390052

L4 layer

L3 layer

L2 layer

L1 layer









119882 = 1205721198821015840+ (1 minus 120572)119882

10158401015840 (2)


119894119895gt 0119886

119894119894=

1119886119894119895= 1119886







1198821015840

1=

[

[

[

[

1 040 032 075

25 1 079 186

3125 1266 1 235

133 0538 0426 1

]

]

]

]




Safetymanagement



Org

aniz

atio

n str

uctu

re an

d re

spon

sibili

ty

Educ

atio

n an

d tr

aini

ng

Safe

ty la

ws a

nd re

gulat

ions

Safe

ty in

vestm

ent

Risk

mon

itorin

g

Pers

onal

bas

ic si

tuat

ion

Perc

eptio

n an

d de

cisio

n er

rors

Skill

-bas

ed er

rors

Team

man

agem

ent

Safe

ty su

perv

ision

ins

pect

ion

and

acce

ptan

ce

Mat

eria

l

Tech

nica

l mea

sure

men

ts

Acci

dent

repo

rt i

nves

tigat

ion

and

treat

men

t

Ope

ratin

g en

viro

nmen

t

Emer

genc

y re

scue

Viol

atio

n op

erat

ions

Mec

hani

cal e

quip

men


11988210158401015840

1=

[

[

[

[

1 1147 1418 2597

0872 1 1238 2262

0705 0808 1 1828

0385 0442 0547 1

]

]

]

]

(3)


1198821=

[

[

[

[

1 062 065 130

161 1 092 198

154 109 1 219

077 051 046 1

]

]

]

]

(4)


1198822=[

[

1 154 469

065 1 263

021 038 1

]

]

(5)


1198823=

[

[

[

[

[

[

1 110 116 880 1173

091 1 105 775 1032

086 095 1 714 950

011 013 014 1 129

009 010 011 078 1

]

]

]

]

]

]

(6)


1198824=

[

[

[

[

[

[

[

[

1 082 092 075 376 847

122 1 111 091 456 1028

109 090 1 083 393 877

133 110 120 1 526 1192

027 022 025 019 1 193

012 010 011 008 052 1

]

]

]

]

]

]

]

]

(7)


1198825=[

[

1 147 114

068 1 078

088 128 1

]

]

(8)


1198826=[

[

1 0947 1632

1056 1 1724

0613 0580 1

]

]

(9)


1198827= [

1 0076

13158 1

] (10)




















1198828= [

1 4325

0231 1

] (11)


1198829= [

1 1212

0825 1

] (12)










6 Conclusions










Acknowledgments


References





































Volume 2014




Journal of











Journal of


Function Spaces






Algebra


































1198641 =


108

= 19759

1198642 =


43

+


65

= 17318

Tau-119910 = 120591119910 = 1198641 minus 11986421198641

=

19759 minus 17318

19759

= 0124

(1)






Material




Emergency rescue


treatment

Skill-based errors




Team management

Risk monitoring


acceptance



0124

0131 0076

00470616

0173 00400189

00550063

00390052

L4 layer

L3 layer

L2 layer

L1 layer









119882 = 1205721198821015840+ (1 minus 120572)119882

10158401015840 (2)


119894119895gt 0119886

119894119894=

1119886119894119895= 1119886







1198821015840

1=

[

[

[

[

1 040 032 075

25 1 079 186

3125 1266 1 235

133 0538 0426 1

]

]

]

]




Safetymanagement



Org

aniz

atio

n str

uctu

re an

d re

spon

sibili

ty

Educ

atio

n an

d tr

aini

ng

Safe

ty la

ws a

nd re

gulat

ions

Safe

ty in

vestm

ent

Risk

mon

itorin

g

Pers

onal

bas

ic si

tuat

ion

Perc

eptio

n an

d de

cisio

n er

rors

Skill

-bas

ed er

rors

Team

man

agem

ent

Safe

ty su

perv

ision

ins

pect

ion

and

acce

ptan

ce

Mat

eria

l

Tech

nica

l mea

sure

men

ts

Acci

dent

repo

rt i

nves

tigat

ion

and

treat

men

t

Ope

ratin

g en

viro

nmen

t

Emer

genc

y re

scue

Viol

atio

n op

erat

ions

Mec

hani

cal e

quip

men


11988210158401015840

1=

[

[

[

[

1 1147 1418 2597

0872 1 1238 2262

0705 0808 1 1828

0385 0442 0547 1

]

]

]

]

(3)


1198821=

[

[

[

[

1 062 065 130

161 1 092 198

154 109 1 219

077 051 046 1

]

]

]

]

(4)


1198822=[

[

1 154 469

065 1 263

021 038 1

]

]

(5)


1198823=

[

[

[

[

[

[

1 110 116 880 1173

091 1 105 775 1032

086 095 1 714 950

011 013 014 1 129

009 010 011 078 1

]

]

]

]

]

]

(6)


1198824=

[

[

[

[

[

[

[

[

1 082 092 075 376 847

122 1 111 091 456 1028

109 090 1 083 393 877

133 110 120 1 526 1192

027 022 025 019 1 193

012 010 011 008 052 1

]

]

]

]

]

]

]

]

(7)


1198825=[

[

1 147 114

068 1 078

088 128 1

]

]

(8)


1198826=[

[

1 0947 1632

1056 1 1724

0613 0580 1

]

]

(9)


1198827= [

1 0076

13158 1

] (10)




















1198828= [

1 4325

0231 1

] (11)


1198829= [

1 1212

0825 1

] (12)










6 Conclusions










Acknowledgments


References





































Volume 2014




Journal of











Journal of


Function Spaces






Algebra











Material




Emergency rescue


treatment

Skill-based errors




Team management

Risk monitoring


acceptance



0124

0131 0076

00470616

0173 00400189

00550063

00390052

L4 layer

L3 layer

L2 layer

L1 layer









119882 = 1205721198821015840+ (1 minus 120572)119882

10158401015840 (2)


119894119895gt 0119886

119894119894=

1119886119894119895= 1119886







1198821015840

1=

[

[

[

[

1 040 032 075

25 1 079 186

3125 1266 1 235

133 0538 0426 1

]

]

]

]




Safetymanagement



Org

aniz

atio

n str

uctu

re an

d re

spon

sibili

ty

Educ

atio

n an

d tr

aini

ng

Safe

ty la

ws a

nd re

gulat

ions

Safe

ty in

vestm

ent

Risk

mon

itorin

g

Pers

onal

bas

ic si

tuat

ion

Perc

eptio

n an

d de

cisio

n er

rors

Skill

-bas

ed er

rors

Team

man

agem

ent

Safe

ty su

perv

ision

ins

pect

ion

and

acce

ptan

ce

Mat

eria

l

Tech

nica

l mea

sure

men

ts

Acci

dent

repo

rt i

nves

tigat

ion

and

treat

men

t

Ope

ratin

g en

viro

nmen

t

Emer

genc

y re

scue

Viol

atio

n op

erat

ions

Mec

hani

cal e

quip

men


11988210158401015840

1=

[

[

[

[

1 1147 1418 2597

0872 1 1238 2262

0705 0808 1 1828

0385 0442 0547 1

]

]

]

]

(3)


1198821=

[

[

[

[

1 062 065 130

161 1 092 198

154 109 1 219

077 051 046 1

]

]

]

]

(4)


1198822=[

[

1 154 469

065 1 263

021 038 1

]

]

(5)


1198823=

[

[

[

[

[

[

1 110 116 880 1173

091 1 105 775 1032

086 095 1 714 950

011 013 014 1 129

009 010 011 078 1

]

]

]

]

]

]

(6)


1198824=

[

[

[

[

[

[

[

[

1 082 092 075 376 847

122 1 111 091 456 1028

109 090 1 083 393 877

133 110 120 1 526 1192

027 022 025 019 1 193

012 010 011 008 052 1

]

]

]

]

]

]

]

]

(7)


1198825=[

[

1 147 114

068 1 078

088 128 1

]

]

(8)


1198826=[

[

1 0947 1632

1056 1 1724

0613 0580 1

]

]

(9)


1198827= [

1 0076

13158 1

] (10)




















1198828= [

1 4325

0231 1

] (11)


1198829= [

1 1212

0825 1

] (12)










6 Conclusions










Acknowledgments


References





































Volume 2014




Journal of











Journal of


Function Spaces






Algebra












Safetymanagement



Org

aniz

atio

n str

uctu

re an

d re

spon

sibili

ty

Educ

atio

n an

d tr

aini

ng

Safe

ty la

ws a

nd re

gulat

ions

Safe

ty in

vestm

ent

Risk

mon

itorin

g

Pers

onal

bas

ic si

tuat

ion

Perc

eptio

n an

d de

cisio

n er

rors

Skill

-bas

ed er

rors

Team

man

agem

ent

Safe

ty su

perv

ision

ins

pect

ion

and

acce

ptan

ce

Mat

eria

l

Tech

nica

l mea

sure

men

ts

Acci

dent

repo

rt i

nves

tigat

ion

and

treat

men

t

Ope

ratin

g en

viro

nmen

t

Emer

genc

y re

scue

Viol

atio

n op

erat

ions

Mec

hani

cal e

quip

men


11988210158401015840

1=

[

[

[

[

1 1147 1418 2597

0872 1 1238 2262

0705 0808 1 1828

0385 0442 0547 1

]

]

]

]

(3)


1198821=

[

[

[

[

1 062 065 130

161 1 092 198

154 109 1 219

077 051 046 1

]

]

]

]

(4)


1198822=[

[

1 154 469

065 1 263

021 038 1

]

]

(5)


1198823=

[

[

[

[

[

[

1 110 116 880 1173

091 1 105 775 1032

086 095 1 714 950

011 013 014 1 129

009 010 011 078 1

]

]

]

]

]

]

(6)


1198824=

[

[

[

[

[

[

[

[

1 082 092 075 376 847

122 1 111 091 456 1028

109 090 1 083 393 877

133 110 120 1 526 1192

027 022 025 019 1 193

012 010 011 008 052 1

]

]

]

]

]

]

]

]

(7)


1198825=[

[

1 147 114

068 1 078

088 128 1

]

]

(8)


1198826=[

[

1 0947 1632

1056 1 1724

0613 0580 1

]

]

(9)


1198827= [

1 0076

13158 1

] (10)




















1198828= [

1 4325

0231 1

] (11)


1198829= [

1 1212

0825 1

] (12)










6 Conclusions










Acknowledgments


References





































Volume 2014




Journal of











Journal of


Function Spaces






Algebra




























1198828= [

1 4325

0231 1

] (11)


1198829= [

1 1212

0825 1

] (12)










6 Conclusions










Acknowledgments


References





































Volume 2014




Journal of











Journal of


Function Spaces






Algebra













6 Conclusions










Acknowledgments


References





































Volume 2014




Journal of











Journal of


Function Spaces






Algebra












































Volume 2014




Journal of











Journal of


Function Spaces






Algebra









Safety Assessment of High-Risk Operations in Hydroelectric ...

Documents

Transcript of Safety Assessment of High-Risk Operations in Hydroelectric ...