PHM - Risk Minimisation [Airforce Institute Presentation]

Air Force Institute of Technology, 23 May 2011

PHM Technology Pty Ltd 1Jacek S. Stecki

The PHMX Technology – Risk minimization

Jacek S. SteckiPHM Technology

X PHM - Prognostics and Health

Management



Key issues – Risk drivers

Reduction of life-cycle costSafety – environmental, personnelReliability – hardware, functionalReduced manning levelsNeed to reduce the volume of scheduled maintenanceSecondary effects of failuresInherent design problemsNeed to reduce spare parts inventoryHigh performance requirementsAvailability of specialised personnelInsurance and classificationCriticality of the equipment to productivity/availabilityCost of lost production or lost availability as a result of equipment failureCost of fixing a problem in terms of repair and bringing the machine back to a serviceable conditionEtc.



Integrated Logistics Support

Integrated logistics support (ILS) is an integrated approach to the management of logistic disciplines in the military

The pupose of ILS is to ensure that the supportability of the system is considered during its design and development in order:

To create systems that last longer and require less support

To reduce costs

To increase return on investments

To assure supportability throught the operational life of the system

The impact of ILS is measured in metrics:

Reliability - Availability - Maintainability (RAM)

Reliability - Availability - Maintainability - Testability (RAMT)

Reliability - Availability - Maintainability - System safety (RAMS).



Integrated Logistics Support

Integrated Logistics

Reliab ility , M ain tainab ility and M aintenance) P lanning

Supply (Spare part) S upport acqu ire resources

Support and T est Equ ipment/Equipm ent

M anpow er and Personnel T raining and T rain ing Support

Technical Data / Pub lications

Computer Resources Support

Facilities Packag ing , Handling ,

Storage, and T ransportation

Design In terface

UK Def ence S tandard (DEFSTAN) 00-600

Supportability of the System

Assuring continued operation and functioning of the systems



Performance-based Logistics (PBL) is an outcome-based, performance-oriented product support strategy

A product support provider (PSP) or product support integrator (PSI) is contracted to meet performance metric (s) for a system or product

The purpose of PBL:

increased system availability, reliabilityshorter maintenance cycles, and/or reduced costs

Thus PBL fits well with ILS

----------------------------------In U.S. Department of Defense (DoD) acquisition programs, the PBL approach is mandated as a first-choice strategy.

– A PBL contract was awarded to Alstom for delivery of trains in France– Also called Performance-based-Contracts

Performance-based Logistics



Reliability - Availability – Maintainability (RAM)

The ability of an item to perform a required function under given conditions for a given time intervalIt is generally assumed that the item is in a state to perform this required function at the beginning of the time intervalGenerally, reliability performance is quantified using appropriate measures. In some applications these measures include an expression of reliability performance as a probability, which is also called reliability.



Risk reduction – CBM/PHM

What is it?Risk assessment using techniques like FMECA, HAZOP, RCM etc.Diagnostics – is the process of determining the state of a component to perform its function(s)Prognostics – is predictive diagnostics which includes determining the remaining life or time span of proper operation of a componentHealth Management – is the capability to make appropriate decisions about maintenance actions based on diagnostics/prognostics information, available resources and operational demand.

D e s ig n

Risk

Sensors

Diagnostic FDI

Prognostics

Failures Identification

Criticality Assessment



PHM - Fusion of the technologies

Sensors Artificial intelligence Neural

nets, fuzzy logic, genetic algorithms

Algorithms (vibration etc.) Communication capabilities Interchange of maintenance

data Integration of data Security of data User friendly interface Autonomy to be provided by

software agents (Jack platform from AOS)

PrognosisPrognosisLayerLayer

Prognostics and Health

ManagementPHM

MaintenanceMaintenanceaware Designaware Design

SensorsSensorsLayerLayer



PHM Paradigm (Joint Strike Fighter F35)

PHM Paradigm

Sensor based Proactive

Prognostic capability

Intelligent Sensors

Data Fusion

Virtual SensingModel-based Prognostics

Maintenance aware Design

Co-current with Design

Optimization

Life Cycle

Autonomous

Open Architecture

Reliable and Robust

Model-based Prognostics



Goals of PHM

Enhance Mission Reliability and Equipment Safety Reduce Maintenance Manpower, Spares, and Repair Costs Eliminate Scheduled Inspections Maximize Lead Time For Maintenance and Parts Procurement Automatically Isolate Faults Provide Real Time Notification of an Upcoming Maintenance Event at all

Levels of the Logistics Chain Catch Potentially Catastrophic Failures Before They Occur Detect Incipient Faults and Monitor Until Just Prior to Failure



Autonomic logistics support (JSF)

"The Joint Strike Fighter should be capable of selfdeploying to any area of operations, worldwide, with a logistics tail comprised of the weapons it will deliver, the fuel needed to fly, minimal support equipment and material, and the personnel needed to generate initial surge operations…"

A highly reliable, maintainable, and intelligent aircraft which encompasses a comprehensive Prognostics and Health Management (PHM) capability to enhance safety, improve efficiency of the logistic chain, and allow scheduling of logistic events to compliment operational planning.

A technologically enabled maintainer who through the use of innovative and automated tools and technical publications will be capable of efficiently and effectively maintaining the JSF with less specialized training.

A fully capable Joint Distributed Information System (JDIS), operating within the GCCS infrastructure, that incorporates advanced information technology to provide decision support tools and an effective communication network linking the JSF with the logistics infrastructure to provide proactive support.

A logistics infrastructure that is sufficiently responsive to support requirements within a timeframe that allows the JSF weapon system to generate the required number of effective sorties at the least cost.



PHM Architecture and Enabling Technologies – JSF Program

Air Vehicle On-BoardHealth Assessment

Health Management,Reporting & Recording

Autonomic Logistics& Off-Board PHM

PVI

MAINTAINERVEHICLE INTERFACE

Mission Critical

PHMData

Displays & ControlsCrashRecorder

Maintenance Interface Panel

IETMsConsumables

On-Board Diagnostics

PMD

.

PMA

In-Flight &Maintenance Data Link

Flight Critical

PHM / Service Info

Database

AMD/PMD

PHM Area Managers

MS Subsys

• Sensor Fusion• Model-Based Reasoning• Tailored Algorithms• Systems Specific Logic / Rules• Feature Extraction

Provides:

• AV-Level Info Management• Intelligent FI• Prognostics/Trends• Auto. Logistics Enabling/Interface

Methods Used:

FCS/UtilitySubsystems

NVMICAWSManager

Hosted in VMC

AVPHM

Hosted in ICP

Structures

MissionSystems

• Decision Support• Troubleshooting and Repair• Condition-Based Maintenance• Efficient Logistics

VS

Propulsion

Results In:

ALIS•Automated Pilot / Maint. Debrief

•Off-Board Prognostics• Intelligent Help Environment

•Store / Distribute PHM Information



Aerospace / Subsea

RisksSevere operating environmentStringent statutory safety standardsSafety critical systemsExpensive MaintenanceLong innovation lead time High technologyConservative attitudesHigh reliability requirementsSingle shot operationsVery high cost of failure

Tools to deal with risksComputer based design methodsReliability and Hazard AnalysisFailure analysis (FMECA/FTA)PHM (Prognostics and Health Management) Condition Monitoring - CBMTesting



CBM/PHM - what are we dealing with?

FMECA

Production Losses

Relia

bilit

y

Condition monitoring

Prognostics

Maintenance

Detection

Diagnosis

Algor

ithm

s

Failure modes

Faults

Simulation

Downtime

TestingRisk M

inim

izatio

n

$$$$$$$!

Training

Fall-back AnalysisHazards

Safety

Training

FMECA

Standards

TrainingFM

ECA

Relia

bilit

yDiagnosis

Sensor fusion

Failure modes

BITTraining

FMECA

Fault Tree

ROI

FMECA

Relia

bilit

y

SensorsDiagnosis

Education

Failure modes

Training

Training

FMECA

Functional

Analysis

Training

Education

Sensor fusionSensor fusion

Artifi

cial

inte

llige

nce

Maintainabil

ity

Availa

bili

ty



Identifying Risk



Reasons for failure of Risk Assessment

Dependencies of failures not identified – spreadsheet vs model basedInadequate Identification of Risks - functional failures (failure modes) vs physical failures Incomplete database of failures (deficient FMECA)Taxonomy – confusion what is the cause, mechanism of failure, fault, symptom and/or failure modeSensor fusion not based on failures dependencies (fall-back – testability) Diagnostic rules not based on dependenciesReliability of Hardware not the same as Functional ReliabilityDifferent models for Criticality and Reliability Assessment



Risk reduction or is it?

Risk is still there if failures are missedWe cannot design a diagnostic system without knowledge of failuresWe do not really know what we should monitorSensors cover only identified failures

D e s ig n

Risk

Sensors

Diagnostic FDI

Prognostics

Failures Identification

Criticality Assessment



BarriersIn many military/industrial applications the metrics for evaluating successful implementation of CBM are not clearly defined (risk, economics, performance)Maintenance requirements/specifications are not defined at the concept formulation stage of the design processIdentification of an optimum level of diagnostic and prognostic requirements and specificationsSelection of an optimum monitoring mix (selection of sensors) which should be system oriented is driven by vendors of sensorsLack of clear guidelines why and when CBM is actually preferred to other maintenance approaches (technical/economic)Skills issues are not addressedMaintenance management systems are inadequateNo knowledge retentionHistorical data, postmortem results not availableUncertainty of ROI (Plant Services Magazine (USA)”..In a survey of 500 companies, less than 3% of respondents were able to achieve a measurable return on their investment in Predictive Maintenance technologies”)



Barriers

The Advanced Technology Program (ATP), of the National Institute of Standards and Technology (NIST), held a workshop on Condition-Based Maintenance (CBM) as part of it's November 17-18, 1998 Fall Meeting in Atlanta.

Discussions with companies identified 3 technical barriers to CBM's widespread implementation: The inability to accurately and reliably predict the remaining useful

life of a machine ( prognostics) The inability to continually monitor a machine (sensing) The inability of maintenance systems to learn and identify impending

failures and recommend what action should be taken (reasoning).

These barriers could potentially be addressed through innovations in three technical areas: Prognostication capabilities Cost effective sensor and monitoring systems Reasoning or expert systems



Risk Assessment FMECA

Failure Modes

Effects

Criticality Analysis

What effect does the failure have ?

Criticality Analysis of failure

Possible Failures FMFMECAECA

FMFMEECCAA

FMEFMECACA



Risk Assessment – e.g. FMECA

Why is FMECA carried outStatutory requirement – must be doneWe need to have audit trail in case of problemsA need to know of how to improve system safetyThe integrator insisted on itReliability people need it

Why FMECA should be carried outWe need to to know what to monitor and what sensors to useWe need to have capability to detect, diagnose and prognose the state of the systemTo design-out failuresWe need to know how the system can fail so we are prepared to deal with itTo enhance diagnostic capabilities



acceptable operating range

Component model

FMEA model

EnergyEnergy

apply forceControlparameterse.g. pressure

Noisee.g.. friction

Measuredvariablee.g.. force

ComponentFunction Definition

High range

Low range

Effect 1 downstream,e.g.. damaged support

Upper limit

Lower limitEffect 2 downstreame.g. failed to lift

PhysicalComponente.g. actuator

Failure Modes and Effects


Component model

Tribological model

WearFriction


LoadVelocityetc.

ComponentRepresentation e.g. drawing

Modeling Failure



Modelling of failure



m

component connector

fault

k

n

Qk down (e.g.. pump leakage)

pressure p down

Qn up (e.g. relief valve open

Qm up (e.g.. check valve leakage)

p

l

Ql down (eg. pipe leakage)

Dependencies Modelling



Fault

Fault

Fault

All faults are enumerated.Transient and steady-state responses to faults are identified

Fault propagation - dependability



Failures - Symptoms/Syndromes

Syn d ro m e o f fa ilu re

Sym p to m

Sym p to m

Sym p to m

Sym p to m

Sym p to mSym p to m



Taxonomy problems

Source - Item Failure term Cause Mechanism Fault L/S FF No Class LIFRAC - failure modesPump, hydraulic Leaking x

Improper flow x

No flow x L/S=Loss/SymptonElectric Motor, AC Winding failure x x

bearing failure x x FF=Functional Failurefails to run, after start x

fails to start x LIF=Lower Indenture Level FailureNSWC - failure modesElectric Motor, AC worn bearing x x

open winding x xshorted winding x xcracked housing x xsheared armature shaft x xcracked rotor laminations x xworn brushes x xworn sleeve bearing x x

Pump, hydraulic Pump cavitation xcomponent corrosion xLow net postive suction head xshaft unbalance x xexternal leakage x xmechanical noise xpositive suction head to low xpump discharge head to high xsuction line clogged xpressure surges xincreased fluid temperature x



Sensor selection



PHM Cycle

PHM requires two main cycles of development, design and operation

The Design Cycle is required in order to generate the knowledge base from which the PHM system can obtain its decisions.

The Operation Cycle describes the steps taken within the PHM system from detection of faults through to conveying instructions or actions.



Interaction between MAD and CBM/PHM Layers at Design Stage

System Concept

System specification

Implementation

Functional diagram

FASTContraints

RiskLayer

PHM Layer

Sensor set

DiagnosticsOptimization

Life cycle

FMECA/HAZOP

Prognostics

Sensors

Techniques

Faults

Techniques

Functions

Manufacturing

PH M L a y e r

M A DL a y e r

D e s ig n p r o c e s s

MAD – Maintenance aware Design



Criteria for RCM Processes

SAE JA1011 “Evaluation Criteria for RCM Processes” defines seven questions for RCM:

What are the functions…of the asset…(functions)? In what ways can it fail…(functional failures)? What causes each functional failure (failure modes)? What happens when each failure occurs (failure effects)? In what way does each failure matter (failure consequences)? What should be done…(proactive tasks and intervals)? What should be done if a suitable proactive task cannot be

found?



MADe software

Fa ilure databa se

Fa ilures de pen dability

B IT des ig n & evaluat ion

Auto Sen sor s elect ion

Fa ilure diagram s

Te stabilit y

Fa ilures critica lit y

Cau sesFa ilure

M ec hanism sFa ults Fa ilure m ode s

Fa ilure ta xon om y

Com p one nt

Sys te m s

Parts

Fa ilure diagram s

Fu nct io nal fa ilure diag ra ms

Auto func tiona l an alysis

Auto qu alitat ive sim ulat ion

Auto report ge nerat io n

Auto de sign ofdiag nos tic ru le s

Fa ilure coverageas ses sm ent

Failure database

Sensor selection/coverage

Coverage of am b ig uity

Datab ase

FM EA/FM ECA

Use d e fine d se nsors



RR250 Engine Lubrication System



Jet Engine Lubrication System Model



Model of pump



Define Component Structure



Define Component Functions



Define Physical Failures



Propagate Functional Failures >> Dependency



Assess Criticality



Produce FMEA/FMECA Report



Assess hardware Reliability



Fault Tree



Define Sensors Locations



Select sensors and generate diagnostic rules



CAD concurrent with MADe



PHM Design Cycle Deliverables

At the end of the risk assessment process, the user has knowledge of:How the system can fail (failure modes)How critical each failure isWhat are the causes of functional failures What are the interactions between functional failuresWhat physical failures are linked to functional failureWhere to place sensors – i.e sensor fusingHow to monitor physical failures How to diagnose functional failureWhat is the expected reliability of the sensing systemWhat is the expected functional and hardware reliability of the system



Despite expectations the acceptance and effectiveness CBM is in question. To be effective:CBM/PHM programs must be designed and executed with the knowledge of the risks to which a system is exposed, i.e. the knowledge how the system failsModel-based failure analysis, defining failures dependencies and improving completeness of risk identifications, should be adopted in preference to spreadsheet and “spreadsheet” like FMECA methodologyModel-based failure analysis should be adopted to enhance knowledge retention, knowledge transfer and to facilitate integration of risk assessment through supply chainsTaxonomies of functions, failure concepts, components should be adopted to improve readability/portability of risk assessment resultsDiagnostic rules and Sensors sets should be selected on the basis of dependencies between failure modes (symptoms >>> syndrome)Clear hierarchy of failure concepts (cause> failure mechanism> fault> failure mode) should be enforced in risk assessment processPhysical failures (cause/failure mechanism/fault) and their symptoms should form basis for BIT design

Concluding Remarks



Plan of Presentation

ExpectationsTechnological progress (sensors, techniques, methods etc.)Barriers to implementationCBM/PHM Risk assessmentModelling failureTaxonomySensor fusionJet Engine lubrication system – exampleConcluding remarks



Preamble

Aircraft and avionics, offshore, marine and other complex engineering systems often operate in harsh environmental and operational conditionsThese applications must meet stringent requirements of reliability, safety, availability and maintainability. To reduce the high cost of development of new products modern design management and a vast array of computer aided techniques are applied during the design and testing stagesMaintainability requirements, long ignored by designers and manufacturers, assumed great importance and forced rethinking of the way the design of new systems should be carried outAs Availability is becoming a major constraint it became important to develop techniques to monitor the health of the system, to diagnose system problems prior to its failure and to prognose the system's remaining lifeEfforts were made to justify these new approaches from an economic point of viewMaintenance Technology became recognized as an academic discipline



Some Expectations

Enhanced system reliability and equipment safetyReduced maintenance manpower, spares and repair costsEliminated scheduled inspectionsMaximised lead time for maintenance and partsProcurementAutomatically isolated faultsReal time notification of an upcoming maintenance event at all levels of the logistics chainCatching potentially catastrophic failures before they occurAbility to detect incipient faults and to monitor them until just prior to failure

based on The Prognostic Requirement for Advanced Sensors and Non-traditional Detection Technologies - A. Hess



Barriers ctd.

CBM programs are initiated without knowledge of how the system can failThe effectiveness of a CBM program cannot be evaluated with current management tools. Education of CBM managers/engineers who could deal with some of these problems is not available at Universities (Monash University was since early 80-ties the only University which had undergraduate/postgraduate programs in CBM supported by multidisciplinary laboratory, the program ceased in 1996). Widespread research in CBM. These endeavors however are invariably directed towards specific techniques (better mousetrap symptom)

Google “Condition based Maintenance” - 33,000,000 hits!Google “Condition based Maintenance barriers” - 1,080,000

hits!.Google “Gearbox Condition monitoring” - 44,000 hitsGoogle “Bearings Condition monitoring” - 350,000 hitsGoogle “Vibration Condition monitoring” - 351,000 hitsGoogle “Contamination Condition monitoring” - 304,000

hits



Analysis of maintenance data from 12 mining sites

12 mine sites – mining trucks, conveyors, shovels etc.Data from mines' maintenance management systemsApprox 500 MB of data collected over period of up to 5 yearsLimited number of detection/diagnostic/ techniquesExternal contractors – no in-house knowledgeOnly 4 sites had useful information – although incomplete

Conclusions (from the report):Sampling/detection and diagnosis do not follow the best practice to achieve meaningful indication of machine stateAny reporting should have have deliverables, or information will not be useful.Unspecified conditions before failure occurredLack of information of how the system, component, part failed ie. postmortemOutline the reactive and pro-active activities.Unknown or missing – grade and quality of roads, drivers, trained, gender, mechanics, conditions, weather, material being hauled, oil used, petrol used, original parts used, shift work, 7 day week, support, underground, humidity, walk around each day, same route etc.Effective FMEA/FMECA Analysis should be conducted prior to monitoring No visible CBM design/planNo possibility to assess ROI



The rise

Rapid progress was made in the 1960s - 1980s in the development of new sensors, symptom monitoring techniques and performance monitoring in aircraft, marine, railways and mining machinery applications. In those years however, monitoring techniques were seldom used together to provide comprehensive and reliable detection and diagnosis of failures. Likewise, research on detection and diagnostic techniques and methodology was usually directed towards a single technique, for example vibration monitoringThe situation changed in the early eighties when the concept of On-condition Maintenance (over the years the name changed to Condition Based Maintenance (CBM) was developed and applied in high risk industries like aviation, mining and offshore oil production. Over the last thirty years there has been huge progress in developing new sensing techniques, diagnostic and prognostic methodologies and in the application of computer analysis techniques



Taxonomy issues

Business

PHM - Risk Minimisation [Airforce Institute Presentation]