reliability-based design, development and sustainment...reliability-based design, development and...

S1SP ARA0127-1

ReliabilityReliability--based Design, Development based Design, Development and and SustainmentSustainment

NDIA’s 2007 T&E Conference: T&E in Support of Operational Suitability, T&E in Support of Operational Suitability,

Effectiveness and Effectiveness and SustainmentSustainment of Deployed of Deployed SystemsSystems

Dr. Anne Hillegas and Dr. Justin Wu, ARA15 March 2007

Expanding the Realm of Possibility

2

Briefing OverviewBriefing Overview

Description of reliability-based methods

Applications and results

Implications of reliability-based methods for T&E

Challenges

Path ahead


3

ARA BUSINESS AREASARA BUSINESS AREASNational DefenseTransportationSecurity Risk & Disaster ManagementGeotechnical & Environmental TechnologiesComputer Software & Supporting Technologies

Probabilistic Function Evaluation System


4

IntroductionIntroductionReliability-based methods are those that

Use the probability of failure as a criterion in the design process

These methods contribute to suitability, effectiveness and sustainability by

Improving system performanceIncreasing operational readinessReducing unnecessary intervention or maintenanceManaging spare parts inventoriesReducing technical and operational risk

Other benefits of these methodsProvide predicted performance across a range of metricsSupport decision-makers by highlighting trade-offs in performance and RAM


5

The “Magic” behind the MethodsThe “Magic” behind the Methods

These methods involveApplying probability distributions to uncertaintiesUsing physics-based modeling to assess the impact of these uncertain factors on system performanceBalancing system design features and inspection intervals against risk


6

Apply State-of-the-Art probabilistic methodsApply State-of-the-Art probabilistic methods

PhysicsPhysics--based Probabilistic Analysisbased Probabilistic Analysis

Develop High Fidelity Computational Models

Geometry

Environment

Material

Optimize:- Risk and Reliability- Maintenance Plan- Test Plan

Optimize:- Risk and Reliability- Maintenance Plan- Test Plan

Uncertainty analysis based on fast probability integration andadvanced random simulations

Other Uncertainties

Compute:- Risk- Reliability- Reliability sensitivities- Response probability

distributions- Failure samples for

maintenance simulation

Compute:- Risk- Reliability- Reliability sensitivities- Response probability

distributions- Failure samples for

maintenance simulation

Model Uncertainties and Variability

Failure Sampling

PRObabilistic Function Evaluation System

© Applied Research Associates, Inc. All Rights Reserved

Most Likely Failure PointJoint probability density

Focus on “Weakest Links” and Most Likely Causes for FailuresFocus on “Weakest Links” and Most Likely Causes for Failures


7

ReliabilityReliability--based Methods that support based Methods that support Suitability, Effectiveness and SustainabilitySuitability, Effectiveness and Sustainability

Reliability-based multidisciplinary optimization (RBMDO)Reliability-based damage tolerance (RBDT)


8

ReliabilityReliability--based Multibased Multi--disciplinary disciplinary Optimization (RBMDO)Optimization (RBMDO)

Optimizes performance subject to multiple reliability-based constraintsIncorporates multi-disciplinary objectives/models (e.g., payload, aerodynamics, shape parameters, weight,…)Accomplishes higher performance over independent optimization of each discipline


9

RBMDO Application RBMDO Application –– Aircraft Wing DesignAircraft Wing Design

NASTRAN model of Advanced Composite Technology (ACT) wingBaseline aircraft: proposed 190-passenger, two-class, transport aircraftCritical Design conditions derived from DC-10-10 and MD-90-30

3804 finite element nodes3770 finite elements (2222 shells + 1548 beams)22 material properties47 shell element properties


10

PerformancePerformance--based objective and reliabilitybased objective and reliability--based constraintsbased constraints

Weight is reduced while reliability is improvedWeight is reduced while reliability is improved

Nor

mal

ized

Wei

ght

Normalized Safety (1/Risk)

0.964

0.9750.987Original

0.90

0.92

0.94

0.96

0.98

1.00

1 10 100


11

Highway Inspections

Another Application Another Application –– Radiation Detector Radiation Detector SustainmentSustainment


12

ReliabilityReliability--Based Damage Tolerance (RBDT) FrameworkBased Damage Tolerance (RBDT) Framework

0

1

Crack Size

PODInitial

Critical Size

At Insp.

Cum

. Pro

babi

lity

Non-Destruc. Inspection Planning

Principal StructuralElement Model

- stress model- life model

Principal StructuralElement Model

- stress model- life model

ProFES

Crack Size

Initial Flaw

Time (Flight hours)

1st Insp.

Fracture Mechanics Damage Tolerance

Update distributionafter maintenance

2nd Insp.

Flaw or FOD Stress

Material Modeling Error

UsageLoad Spectra

Inspection time

Time

Fully Integrated Finite Element stress, Fracture Mechanics life and ProFES analyses

Failure Sampling

Joint probability density

Most Likely Failure Point

Failure Samples

Flight Hours

Prob

. of F

ract

ure

Without InspectionWith Inspection

Max. risk reduction


13

Project sponsored by FAACritical structures must maintain very small probability of failureSupplement current “safe-life” design approach (which tends to be too conservative)Systematically treat variability/uncertainty in:

usage, load, flaw, material, geometry, modeling error, defect detection capability

Maintenance planning for:Non-Destructive Inspection, inspection frequencies, repair/replacement

Has wide applicability to structures with material or manufacturing flaws

Select appropriate NDI intervalOptimize sustainment strategies

ReliabilityReliability--Based Damage Tolerance (RBDT) Methodology Based Damage Tolerance (RBDT) Methodology for Rotorcraft Structuresfor Rotorcraft Structures

DistributionsBounds or Safety Factors

Other Variables

ReliabilitySafety marginSafety Measure

Optimized schedules for max. risk reduction

Life/No. of inspections

Inspection Schedule

0 < Probability < 1 Certain (Safe-Life)Flaw Existence

Probabilistic distributionA given crack sizeFlaw/Defect size

Probability & confidenceBounds or Safety Factors

Underlying Principles

ProbabilisticDeterministic

Flight Hours

Prob

. of F

ract

ure Without Inspection

With Inspection Max. risk reduction


14

RBDT Application RBDT Application ---- Rotorcraft Spindle Lug ModelRotorcraft Spindle Lug Model

Goal Goal –– Optimize inspection schedule to minimize riskOptimize inspection schedule to minimize risk


15

RBDT Yields Optimal Inspection ScheduleRBDT Yields Optimal Inspection Schedule

Service Insp Crit. parts

Reducible Risk

[ ( ) ( )] [ ( )]C D

o of f

p p

p t p t E POD a

= ⋅

= − ⋅

Systematic approach for probabilistic fracture mechanics damage tolerance analysis with maintenance planning under various uncertainties

Stage 1: compute risk without inspectionStage 2: compute risk with inspection by simulating inspection and maintenance effects using the samples generated from the Stage 1 failure domain


16

ReliabilityReliability--based Damage Tolerance Applicationbased Damage Tolerance ApplicationPipeline Maintenance OptimizationPipeline Maintenance Optimization

Corrosive environments cause metal lossesTwo major failure modes with uncertainties

Burst can cause a high consequence but is less likely to occurLeak has a low consequence but is more likely to occur

In-line inspection devices can detect significant defects Options to maintain integrity at different risk/cost:

Defect monitoring using high resolution ILI devices (cost issue)Repair/replace sections (cost issue)Reduce operating pressure (production loss)Corrosion mitigation (cost issue)

ObjectiveDevelop maintenance optimization software using ILI data and probabilistic failure and cost models

In Line InspectionMagnetic Flux Leakage and other ILI

devices can travel through pipelines to detect metal losses


170 2 4 6 8 10 12 14 16 18 20

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1Pf at 20 Yrs vs. Inspection Time

Pf a

t 20

Yrs

Next Inspection Time (yrs)

Optimal Inspection Time at 3 Yrs

Prob. Defect Model

Prob. Defect Model

ILI DataILI DataProb.

CorrosionModel

Prob. Corrosion

ModelRisk/Reli-Based Decision Model

Risk/Reli-Based Decision Model

Maintenance Options

Maintenance Options

Opt. Maintenance• Repair/No Repair• Repair/Replace• Repair Methods• Future InspectionSchedules

Opt. Maintenance• Repair/No Repair• Repair/Replace• Repair Methods• Future InspectionSchedules

ILI Meas. Error

ILI Meas. Error

Prob. Burst/Leak

Models

Prob. Burst/Leak

Models

Prior Operating Conditions

Prior Operating Conditions

Projected Operating Conditions

Projected Operating Conditions

Cost ModelCost

Model

Risk is minimized if the next inspection and repair time is 3 years from the last inspection

Pipeline Maintenance Example Pipeline Maintenance Example

Pipeline Maintenance Pipeline Maintenance Optimization MethodologyOptimization Methodology


18

Another Application Another Application –– Robotic/Unmanned Systems Robotic/Unmanned Systems DesignDesign


19

Implications for T&EImplications for T&E

Material Constitutive Modeling

Component Analyses

Subassembly Analyses

Global Analyses

Material Testing

Observables

2.5

2.0

1.5

1.0

0.5

0.0

LON

GIT

UD

INA

L S

TRE

SS

(G

Pa)

0.030.020.010.00-0.01-0.02TRANSVERSE STRAIN LONGITUDINA

0o

0o

15o

15o

22.5o

22.5o

30o

30o

45o

45o

60o

60o

90o

Experimentaldata

0

0.5

1

1.5

2

2.5

3

3.5

0 0.01 0.02 0.03 0.04 0.05

ExperimentCalculation

StrainS

tress

(GP

a)

Validation Tests

Building block approach Building block approach requires testing at all levelsrequires testing at all levels


20

WrapWrap--upup

AccomplishmentsTools for taking a reliability-based approach to design and sustainment that can result in cost savings and risk reductionTraceable predictions of reliability and performance changesComputationally fast and efficient methods

ChallengesScalability of approach Understanding material failure propertiesCascade of variable and interrelationships

Non-unitized structures“User friendliness”

Enable use by decision-makers


21

Path AheadPath Ahead

Mature/prove methods for specifying probability distributions

ARA’s Klein Associates Division (cognitive scientists)

Refine the understanding of the impact and interaction of the uncontrollable random variablesAccelerate collection of data to inform physics-based modelsContinue to evolve cost- and time-effective testing approaches for verifying reliability

reliability-based design, development and sustainment...reliability-based design, development and...

Documents