Determining a defensible preventive maintenance plan

Determining a defensible preventive maintenance planPresented by Jim Kennedy, CPEng, CFAM, CAMA

Interlogis Consulting

1 August 2017 Interlogis Consulting

Course Agenda

Day 1

• Introduction

• Session 1 – Maintenance and its management

• Session 2 – Risk and Reliability

• Session 3 – RCM and Condition Monitoring task period


Course Purpose and Outcomes

• Purpose

• To prepare for determining an objective condition monitoring program using

optimising and verifying algorithms in a defensible manner.

• Outcomes

• Define the defensible budget concept

• Understand and list the maintenance management objectives in Asset

Intensive industries

• Define maintenance and identify the different types

• Describe the failure characteristics associated with preventive maintenance

plans

• Explain the role and development of Preventive Maintenance Plans

• Explain the basic process of FME(C)A/RCM/LORA/TA

• Determine condition monitoring task periods

• Verify condition monitoring task period

Session 1Maintenance and its ManagementSetting the scene


The defensible (maintenance) budget

Assures agreed and verifiable objectives of:

• Safety and environmental risks managed

• Required performance achieved at known level of assurance

• All done at a desired balance between the performance, the cost and the

residual risk

Defensible is defined as comprising solutions that are:

• Fact and risk based

• Fully traceable to system/asset output requirements

• Demonstrably good practice (international and national standards)

• Compliant with statutory and regulatory imperatives

• Implemented by competent (certified) staff

• Supported by verified technology (information and decision systems)

• Transparently and verifiably costed

• Deliverable in the agreed time frame


What is Maintenance?

All activities necessary to retain an item in or return it to a serviceable condition.

Blanchard 1974

Nowlan and Heap 1978

IEC International Electrotechnical Vocabulary*


Maintenance objectives - aerospace industry

• Preserve inherent levels of safety and reliability designed into equipment

• Restore safety and reliability to their inherent level when deterioration has

occurred

• Obtain the information to improve all processes associated with the

system lifecycle

• Do the above at minimum cost of ownership

Adapted from

Nowlan and Heap page xvi

December 1978

Activity 1.3


Maintenance Terminology

Maintenance

ObjectivesPreventive

Maintenance

Corrective

Maintenance

Condition Monitor

Hard Time Activity

Functional Test

Repairs (Stds)

Renewals (Cost)

Unplanned

Planned


Technical Maintenance Plans

• Preventive maintenance policies developed using FMECA/RCM to

achieve inherent asset reliability.

• TMPs were introduced in the Royal Australian Air Force in 1970s, Rail

Industry in 1980s. Power Industry in 1990s. Electrical distribution in 2000s

• Policies cover:

• Which assets are to be maintained?

• What maintenance is to be done?

• When the maintenance is to be done?

• How the maintenance is to be done?

• Where the maintenance is to be done?

• Who is authorised to do the maintenance?


Maintenance Requirements Analysis

Maintenance requirements analysis process

FMECA TA

Failure Mode Effects and Criticality Analysis – FMECA

Reliability Centered Maintenance – RCM

Level of Repair Analysis - LORA

Task Analysis - TA

TMP

Service

SchedulePreventive

CorrectiveRepairsAssets

Business

Functions

Failure modes and parts

Failure modes parts,

risk and causes

LORA

RCM

Session 2Risk and ReliabilityDoing it by the numbers


Risk management process

Co

mm

un

icat

e an

d C

on

sult

Mo

nit

or

and

Rev

iew

Establish the Context

Identify Risks

Analyse Risks

Evaluate Risks

Treat Risks

AS HB89 Risk assessment methods

Event Tree Analysis

Fault Tree Analysis

Cause and Consequence

Bow Tie Diagram

Failure Modes and Effects Analysis

(FMEA)

Fault Mode, Effects and Criticality

Analysis (FMECA)

Reliability Block Diagram

Human Reliability Analysis

Consequence/Likelihood Matrix

Cost Benefit Analysis

Multi Criteria Decision


FMECA – a risk process

RISK

FunctionalPerformance

BusinessRequirement

AssetSolutions

SystemFunctions

EquipmentFunctions

Failure Modes

Failure Effects

Failure Probability

H

L

Consequence

Identify Mitigation

Preventive Actions

Corrective Actions

Operator Actions

Maintainer Actions

Redesign Actions

Functional Block Diagram

Reliability Block Diagram


Reliability and the “Bathtub” Curve

Reliability Definition

The probability that a defined item shall operate to a defined standard for a

defined period of time in a defined operating environment

Wear-in Wear-outUseful life

H(f)

Age

Hazard

Function


Hazard function in reliability

The hazard function defines:

• Conditional probability of failure in a particular time interval which can be

dependent on previous intervals

• i.e. the expected future (T+Δt) rate of failure of an item of equipment given

that it has survived to a particular age (T)

• T may be measured in time or events.


Weibull Family of Curves

Probability Density Function Reliability Function Hazard Function

Raw failure data set Likelihood of survivingto time ‘t’

Likelihood of failurehaving survivedto time ‘t’

MTBF


Relating characteristic to task type

Nowlan and Heap

AD AO66579

Source NASA

88%

pd

f

time

% 1968

4

2

5

7

14

68


RCM and technology impact

3

1

7

3

6

% 2001

2

10

17

9

56

6

% 1968

4

2

5

7

14

68

Nowlan and Heap

AD AO66579

Reliability Centered Maintenance Source NASA

89% 77% 71%

% 1982

42

29


pd

f

time

Probability density function (Constant failure rate)

Number of failures of survivors in a discrete time period

f(t)= l e -lt ie if 100 items at start and failure rate of 10% then0 = 100 1 = 90 (100 - 100/10 = 90)2 = 81 (90 - 9/10 = 81)3 = 73 (81 - 81/10 = 73)4 = 66 (73 - 73/10 = 66)


time

pd

f

Random failure pattern - Condition monitor

Each Item achieves maximum life

but at the cost of many Condition

Monitoring tests

100%

0% Time

Resistance to

failure

100%

0% Time

Resistance to failure

Conditional Defect Point

Degrading Asset Condition


Random failure pattern

Number of survivors at t = MTBF


RCM – 7 Questions and 4 Answers

1. Which assets are important to the business?

2. What are its functions?

3. How does it fail to perform that function?

4. What causes it to fail?

5. What happens when it fails?

6. How can that failure be managed?

7. What can be done if the failure cannot be managed?


RCM - The four risk based solutions

Examine condition to detect

potential failures (Condition

Monitor)

Restore or discard before a

maximum age (Hard Time)

Check to find failures that are

not evident (Failure Finding)

Apply default tasks of “run to

failure” or “redesign”


Knowledge outcomes of MRA

The analysis achieves a detailed listing of:

• what functions and equipment comprise the rail system solution

• what equipment related failure modes adversely impact function

• what are the risks associated with those failure modes

• what preventive tasks (controls) will reduce failure risk

• what corrective tasks are available to recover from failures

• what quality elements are necessary to assure task effectiveness

• what hazards are Inherent in the tasks

• what controls are necessary to manage those hazards

Session 3CM Determination and verificationWhat is my inspection period?


Seven Step Analysis Process

1. Breakdown the asset into systems and items of equipment.

2. Prioritise the assets for analysis according to risk exposure from failure.

3. Collect system information and define each failure problem to be

addressed.

4. Establish possible preventive maintenance strategies for dealing with

each failure cause based on its consequence.

5. Evaluate the validity of each particular preventive maintenance policy

(task and frequency).

6. Determine what to do if there are no applicable and effective

maintenance policies.

7. Package the valid preventive maintenance policies into cost effective

schedules.


Preventive maintenance task options

• Condition monitoring is intended to identify the

potential for item failure with sufficient warning to allow

maintenance prior to failure.

• The MIL-STD-2173 algorithm determines the optimum

number of examinations across the CF interval.

• CSIRO Maths Division - validation paper (Mar 2001)

• Additional consideration should be given to critical

operating equipment that is not readily removed from

service for maintenance (requirement for warning time)

Is a condition

monitoring task

applicable and

effective

Is a scheduled

restoration task

applicable and

effective

Is a scheduled

discard task

applicable and

effective

Redesign or

Run to failure

applicable and

Is a failure

finding task

effective


Matching condition to task period

100%

0% Time

Resistance

to failure


Standards

Decision


Functional

Failure Point

Task Period < Warning period and

Task Effectiveness of 0.95

Warning

Period

19 Conditional Failures

1 Functional Failure20 Items

You cannot stop the failure – but!

You can change the consequence

//localhost/Users/jameskennedy/Documents/Interlogis Clients/State Water/Validation of CF Optimistion Method .xlsx

//localhost/Users/jameskennedy/Documents/Interlogis Clients/State Water/Validation of CF Optimistion Method .xlsx


CM Task Optimisation formula – MIL-STD-2173AS

n =

ln

-MTBF T

* Ci

(Cnpm – Cpf)*ln(1-)

ln(1-


Impact of task effectiveness

How many times do I need to do a 50% effective task to achieve a 95% outcome

0.50 0.25

0.75

0.12

5

0.062

50.03125

0.875

0.9375

0.96875


Examples of human error/violation rates - Reason

Task Error Scenario 1

• Unfamiliar activity

• Performed at speed

• No idea of consequence


• Routine

• Highly practiced

• Rapid delivery

• Low skill


• Very familiar

• Well designed

• Highly practiced routine

• Trained and motivated deliverer

• Time available to correct errors

Violation Type 1• Compliance unimportant

• Easy to violate

• Low likelihood of detection

Violation Type 2• Compliance important (legal)

• Low chance of detection

Violation Type 3• Culturally unacceptable

• Low likelihood of detection

• Low likelihood penalty

12

150

12500

13

130

1120


Sample CM Assessments

$0

$20,000

$40,000

$60,000

$80,000

$100,000

$120,000

$140,000

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Co

st

of

Fu

ncti

on

Examination Period Mths

Cost of Examination/year

Cost of Conditional failures/year

Cost of Functional Failures/year

Total cost

Cost of Examination/year 1

Cost of Conditional failures/year 1

Cost of Functional Failures/year 1

Total cost 1

Variables 0

Costofexamination $40

MTBFmths 60

Taskeffectiveness% 95%

CostofFailure(Cond) $1,500

CostofFailure(Funct) $50,000

Population 100

Warningperiodmths 12

OptumimNumberofexams 2.20

OptimumPeriodMths 5.46

Days 166

CostofExaminations/Yr $8,797

CostofConditinalFailures/yr $29,959

CostofFunctionalFailures/Yr $1,507

TotalCost/Year $40,262

NumberofFunctionalFailures/Yr 0.03

NumberofConditionalFailures/Yr 19.97Failures/Yr 20.00

Activity 1.1


Planned maintenance opportunities

0% Time




20x20 Items

380 Conditional Failures (95%)


20 Items

1.6 x Task period = Warning period

Warning Period

8 Items


Caveats and assumptions

The warning period (T) is consistent

• If highly variable, constant auto monitoring and shutdown

Success probability is consistent over T

• Conservative assumption as success improves closer to failure

Warning period (T) is less than MTBF/5

• If T becomes larger then process approaches wear out

Success probability is less than 1

• Examples of extremes (70% to 99%)

Candidate arrivals across T are random

• Fewer candidates as failure approached

0% Time




20x20 Items



20 Items

1.6 x Task period = Warning period

Warning Period

8 Items


Condition Monitoring methods matrix

Source – NASA RCM Guide


Selecting condition monitoring process

Pole mounted substations – what is most cost effective approach.

Hands on examinationUse of thermography


Condition Monitoring – Verifying your task estimate

Figure 1

CF Verification Graph CM Optimisation Graph

(CF interval Vs Exam success) (annual exam cost Vs task interval)

Activity 1.2

Session 4Failure FindingWhat is may failure finding period?


Failure finding task frequency

Primary Item

Protective Item

AdverseEvent

LOC

PossibleOutcomes

Model

Optimum Task Frequency

MaintenanceCost

Cost ofFailures

Total Cost of Task Freq

$

Cost Profile

Is a condition

monitoring task

applicable and

effective

Is a scheduled

restoration task

applicable and

effective

Is a scheduled

discard task

applicable and

effective

Redesign or

Run to failure

applicable and

Is a failure

finding task

effective


Functional check sample outcome – Method 1

MTBFFmths 120 =1in 1,861 peryear

Q 0.0327839

LOCEProb 0.000537441

TestMths 4

MTBFF 120

UAo 0.01639344

Additional 0

Protection


Failure Finding model – Method 2

MultipleFailure

Primary

Protective Failed

Check Period X+1 Check Period X+3

TimelineSystem

Check Task

Failed

Check Task

Check Period X+2

Check Task

FailFixFail

Time between failures

Fix

Task Effectiveness


FET Failure finding model – Method 2

Possible

Outcomes

or

Primary Item

Protective Item LOCE(Energy)

Low Impact - high probability

Mid Impact - mid probability

High Impact - low probability

and


Sample outcome – Method 2

Fault-EventTreePrimaryMTBF(mths) 120

ProtectiveMTBF(mths) 120

NumberofProtectiveElements/Primary 1

CostofFunctionalTest $200CostofPrimarySystemRepair $2,000

CostofProtectiveElementRepair $1,000

ProbabilisticCostofaFunctionalFailure $200,000PopulationofPrimaryElements 50

Thank you for your [email protected]

Determining a defensible preventive maintenance plan

Documents

Transcript of Determining a defensible preventive maintenance plan