producing effective maintenance strategies to control

Resilience Engineering Research Group

Producing effective maintenance strategies to control railway risk

ClaudiaFecarottiandJohnAndrewsResilienceEngineeringResearchGroup

TheUniversityofNottingham

PSAM14–ProbabilisticSafetyAssessmentandManagement

16-21September2018,UCLAMeyer&ReneeLuskinConferenceCenter,LA


Motivations

•  Complexanddiverseportfolioofassets•  Heterogeneity•  Highlyinterconnected(dependencies)•  Expensivetomanage

•  Manyassetsaresafetycritical

•  Maintenanceisvitaltocontroltheriskandmaintainhighlevelsofservice


Motivations

Need for an effective approach to asset management to run a SAFE,RELIABLEandAFFORDABLErailway

•  Bespokemodelstopredictassetsresponsetomaintenance•  Whole-life/Whole-systemapproach•  Linkassetmaintenancetosystemperformanceandsafety


Railway Asset Management Framework

RailwayAssetManagementModellingFramework tosupportboth localisedandsystemicoptimaldecisionsoninfrastructuremaintenance

Libraryofmodels

•  Statisticalmodels•  Predictivemodels•  Optimisationmodels

Decisionlevels

•  Infrastructure:asset/section/route/network•  Planningstage:strategic/tactical/operational

Procedurestolinkmodelsforwhole-systemrepresentationandanalysis

Pastdata

Predictivemodels

Futuredata

Optimisationmodels

Optimaldecisions


Predictive models

•  Assetstatemodelstoassessassetsresponsetomaintenance

•  Serviceprovisionmodeltoevaluatedelaysandjourneycancellations

•  Riskandsafetymodelstoevaluateriskandconsequencesofhazardousevents

EnablepredictionofKPIs

•  Assetsconditions•  Servicereliability•  Safety•  Costs


Asset State Models DEGRADATIONANDFAILURE

INTERVENTIONSTRATEGIES:•  Inspectiontypeandfrequency•  Levelsofdegradationtriggeringintervention• Componentsreplacementpriortofailure

• Condition• Age• Usage

• Opportunisticmaintenance• Renewal• Enhancement• Resourcesavailability(equipmentandpersonnel)

For any asset management strategy predictdistributionsof:•  Assetconditions

•  Failuremodesprobabilities•  Durationoffailed/degradedstates•  Futureconditions(atanytime)

•  Assetavailability•  Numberofeachinterventiontype•  Assetremaininglife

&

ASSESSANDCOMPAREDIFFERENTMAINTENANCESTRATEGIESTOSUPPORTDECISIONS


Common modelling technique: Petri nets

•  State-based•  Stochastic•  Simulationfriendly•  Anydistributionoffailuretimes

•  Assetswear-out(increasingfailurerate!earlyreplacementoption)•  Dependencyonpastconditions/events

•  Complexmaintenanceprocesses:•  Servicing,inspection,replacementpriortofailure(basedon;condition,age,use),reactiverepair,refurbishment,renewal

•  Conditionandriskbasedinspection•  Concisestructurecomparedtothecrediblealternatives•  Modularity(Easylinkingtoformthesystemmodel)•  Distributionofoutputsratherthanpointestimates


Petri nets

PLACEcomponentstate,physicalcondition,logicalcondition

TRANSITIONevent:degradation,failure,repair(immediate,timed-deterministicandstochastic)

TOKEN(numberoftokensineachplacedeterminethestateofthesystematanytime–MARKING=SYSTEMSTATE)Tokensare“consumed”and“produced”wheneventsoccur(transitions“fire”)determininganewsystemstate

W W WF F F

D D D

R

Initialworkingstate Failureoccurs–transitionDfires

Componentisrepaired–transitionRfires

RR


Track geometry maintenance model

•  4degradedstates

•  Degradedstateaffectingserviceandsafety:•  speedrestriction•  lineclosure

•  Weibulldistributionoftimestodegrade

•  Notasgood-asnewaftermaintenance

Places(conditions):Progressivelevelsofdegradationtriggeringdifferentmaintenanceinterventions

Transitions(events):Degradationbetweenconsecutivestates(Anydistributionoftimestodegrade)



•  4degradedstates•  Degradedstateaffectingserviceandsafety:

•  speedrestriction•  lineclosure

•  Weibulldistributionoftimestodegrade•  Notasgood-asnewaftermaintenance

•  Periodicinspection

•  Currentstatesrevealed



•  4degradedstates•  Degradedstateaffectingserviceandsafety:

•  speedrestriction•  lineclosure

•  Weibulldistributionoftimestodegrade•  Notasgood-asnewaftermaintenance


•  Currentstatesrevealed:

•  UnrevealedneedforSRandLC(safety)

•  RevealedneedforSRandLC(service)



•  4degradedstates

•  Weibulldistributionoftimestodegrade

•  Notasgood-asnewaftermaintenance


•  UnrevealedneedforSRandLC(safety)

•  RevealedneedforSRandLC(service)

•  Revealedstatestriggermaintenance

•  Maintenanceeffectiveness

•  Side-effectoftamping

•  Renewalstrategies(e.g.age,pastmaintenance)


Track geometry maintenance application

Strategy Inspectionperiod(T14)

Meantimetoperformroutinemaintenance

(T10)

Meantimetoperformmaintenancefromspeed

restriction(T11)

Meantimetoperformimmediaterepair

(T12) θ(days) µ(days) σ2(days2) µ(days) σ2(days2) µ(days) σ2(days2)

1 15 20 5 5 1 1 0.1 2 15 20 5 10 2 1 0.1 3 15 30 5 5 1 1 0.1 4 15 30 5 10 2 1 0.1 5 15 40 10 5 1 1 0.1 6 15 40 10 10 2 1 0.1 7 120 20 5 5 1 1 0.1 8 120 20 5 10 2 1 0.1 9 120 30 5 5 1 1 0.1 10 120 30 5 10 2 1 0.1 11 120 40 10 5 1 1 0.1 12 120 40 10 10 2 1 0.1

Table 3 Maintenance strategies.

Whatdoweuseitfor?Toinvestigateassetresponsetomaintenance.

SDop SDrm SDsr SDlc opportunistic

maintenanceispossible(associatedtoplaceP2)

routinemaintenanceisrequired

(associatedtoplaceP3)

SRandemergencyrepairrequired

(associatedtoplaceP4)

LCandimmediaterepairrequired

(associatedtoplaceP5) 1.5 1.8 2.5 3.5

Table 1 SD threshold values for each degraded state

T1 T2 T3 T4 T5 β η β η β η β η β η 1.5 600 1.5 500 1.6 370 1.7 280 1.8 740

Table 2 Weibull parameters associated to each stochastic transition representing degradation


Track geometry maintenance results

15

16

17

18

19

20

21

1 45

89

133

177

221

265

309

353

397

441

485

529

573

617

661

705

749

793

837

881

925

969

Num

berofinterventio

ns

Number of interventions per lifetime averaged over the number of simulations

S1

S2

S3

S4

S5

S6

S7

S8

S9

S10

S11

S12

Numberofsimulations

1414.515

15.516

16.517

17.518

18.519

1 2 3 4 5 6 7 8 9 10 11 12

strategies

NumberofRMinterventions

00.10.20.30.40.50.60.70.80.9

1 2 3 4 5 6 7 8 9 10 11 12

strategies

NumberofSRimposed

0.86

0.88

0.9

0.92

0.94

0.96

0.98

1 2 3 4 5 6 7 8 9 10 11 12strategies

Probabilityofbeeingingoodconditions

0

0.0001

0.0002

0.0003

0.0004

0.0005

0.0006

1 2 3 4 5 6 7 8 9 10 11 12

Strategies

ProbabilityofimposedSR

0

0.0002

0.0004

0.0006

0.0008

0.001

0.0012

0.0014

0.0016

0.0018

0.002

1 2 3 4 5 6 7 8 9 10 11 12

strategies

ProbabilityofunrevealedneedforSR


Conclusions and future work

•  Needforasystematicapproachtorailwayassetmanagement:

RailwayAssetManagementModellingFramework

•  Assetstatemodelstopredictassetresponsetomaintenance

•  Modellingtechnique:Petrinet

•  Results(failuremodesprobabilities)areinputtoserviceandsafetymodelsthusenablingtolinkmaintenancetosystemperformanceandsafety

Futurework•  Continuepopulatingtheframework…..

producing effective maintenance strategies to control

Documents