1

Reliability-centered
Knowledge
Presented by
Murray Wiseman

Optimal Maintenance Decisions Inc.

To

Cerrejón Coal
Sept-Oct, 2005

Reliability-centered Knowledge
• Part 1: Knowledge management
• Part 2: Condition based maintenance
• Part 3: Reliability-centered maintenance

The solution providers

Consultants

Original Equipment
Manuafacturers

Maintenance contractors

Maintenance
technology vendors

13

1
 UML class diagram

WorkOrder
is
does

15

UML class diagram

WorkOrder
-workOrderNo : Integer
-equipment : String
-workingAge : Double
+calculateCost() : Double

13

The use of recorded information

to improve OEE, reliability, 1. Adjust PM task

safely, at lowest cost intervals & PM tasks
by assessing:
2. Assess desirability of
1. Failure Data additional PM tasks
In order to
2. Inspection findings 3. Eliminate unnecessary
(over intensive) PM
tasks
This is a continuous process! 4. Improve failure
response
5. Redesign and Design

15

2
 UML class diagram
WorkOrder
-functionCompromised : String
-failure : String
-cause : String
-effects : String
-consequences : String

15

Requirements of maintenance
information
How much is enough?

19

The functions of a CMMS knowledge base

1. To determine the types of failures the equipment is actually
exposed to as well as their frequencies
2. To expose the consequences of each failure, ranging from direct
safety hazards through serious operational consequences, high
repair costs, long out-of-service times for repair, to a deferred need
to correct inexpensive functional failures
3. To confirm that functions originally classified as evident (during
RCM analysis) are in fact evident to operating personnel during the
normal performance of duties

4. To identify the circumstances of failure in order to determine whether the

failure occurred during normal operation or was due to some external
factor (accidental damage)
5. To confirm that on-condition (CBM) inspections are really measuring the
reduction in resistance to a particular failure mode
6. To inform us of the actual rates of reduction in failure resistance in order
that we may determine optimum inspection intervals

19

3
 The Purpose of a Reliability-Centered Knowledge Base

7. To record the mechanism involved in certain failure modes in order to

identify new forms of on-condition inspection (CBM) or parts that
require design improvement, or improve diagnostic response.
8. To identify those tasks assigned originally as default actions but that
do not prove applicable and effective
9. To identify maintenance packages that are generating few trouble
reports and are candidates for longer interval schedules
10. To identify items that are not generating trouble reports
11. To record the working ages of assets and components at which
failures occur

In summary, the purpose of a Reliability-Centered Knowledge Base is to use all

of the above to
IMPROVE ASSET OEE AT LOWEST COST, SAFELY

19

Proposed Objective

To build and use a reliabilty-

centered knowledge solution

19

The UML Context Diagram

Unified Modeling
Language

20

4
 Data model

20

Child work order

22

Use Case Diagram - Complete

the work order form

23

5
 Additional attributes

24

The UML use case diagram

1. Simplified guidelines and training

document,
2. Accessible examples,
3. Supervision, discussion
4. Support,
5. Evolving failure codes ,
6. Revision and Audit capabilty

Sequence
diagram
25

Auditing for quality of

knowledge base

26

6
 UML sequence diagram

Using CBM
data 27

Anticipated vs. Actual Experience

• Compared failure modes in FMEA to those
encountered in the field
– Turbofan Engine
Extracted from Failure modes
CaseBank study “anticipated”
published in IEEE
Paper “Comparison of
FMEA and Field 610
Experience for a
Turbofan Engine with
FMEA Overlap of 142 Failure modes
Application to Case-
Based Reasoning”
727 Failure modes
Field “experienced”
Experience (LRUs primarily)

32

Using maintenance data

Analyzing data for reliability

improvement

7
 Using Maintenance Data

Inspection Data Examples

33

Average availability from a safety device

inspection inspection
1
Probability of survival without failure

100% + 78%
= 89%
.78 2

.61

.50 .47
.37
.29
.22
Exponential survival

0
0.25 0.50 0.75 1 1.25 1.50
Ratio of interval to mean time between failures38

Refining the maintenance program

Analysis of Response to
results of unanticipated
scheduled tasks failures
default decisions
made in the absence
of information

44

8
Conditional probabilty of failure for 200

0.4

0.3

0.2
hour intervals

Total removals
0.1 Potential failures
Functional failures
0 1000 2000 3000 4000
Operating age since last shop visit (flight hours)
45
Conditional probability of failure

Total removals

Unverified failures Verified failures

Failure mode C

Failure mode B
Infant
mortality Failure mode A

Operating age
46

Weibull analysis

47

9
 Weibull analysis

49

CBM Effectiveness Comparison

CBM effectiveness is related, ultimately, to how "good" the condition data is.

50

Significant components

54

10
 Events

56

Acquiring Maintenance
Information
Using the EXAKT Work Order
Processor

58

Simplified but typical work order form

58

11
 A work order form that accomodates the 16
reliability information elements

61

Event type
• FF - the ending and renewal of a component (failure mode) due to a
functional failure
• PF - the ending and renewal of a component (failure mode) due to having
detected a potential failure in time to avoid the more dire consequences of a
FF.
• S - the ending and renewal of a component (failure mode) for any reason
other than (functional or potential) failure. (For example preventively
replacing the component.)
• B - the beginning of the life of a component in the item (if not FF, PF, or S)
• BSA - the beginning of a period of temporary removal (suspended
animation) of a component from the item.
• ESA - the return of the same component to the item after a period of
suspended animation
• SA - the beginning and ending of a period of suspended animation if
reported on the same work order.
• MR - the minor repair of the item. It does not renew any components.
Sometimes it will impact the monitored data. For example, a calibration, a
shaft alignment, an oil change, the balancing of an impeller, and so on.

63

EWOP generated Events table

65

12
 EWOP generated RCM records

66

Uniqueness of a work order

• item
• function
• failure
• cause

66

Functional failure and renewal

Events

Example 1

67

13
 Potential failure

Example 2

68

Suspension

Example 3

68

New installation

Example 4

69

14
 Meter reset

Example 5

69

Installing a used component

Example 6

70-71

Suspended animation

Example 6

71

15
6

Deciding on CBM

Why? When?
Preferred policy for 3 reasons

83

The basic model

Basic assumptions:

1. The potential failure is reliably

detectible
2. The P-F interval (the time between
detectible potential failure and
functional failure) can be
estimated, and
3. The inspection interval has been
set to one half the P-F interval,
and
4. The inspection interval has been
ascertained to allow adequate
warning time in which to react
appropriately to the potential
failure.

84

Decision algorithm

86

16
 1
Analysis
2 3 4
Decision
5 6, 7
Resourcing

F Failed Cause Effects Conseq Task •Training

E • Training
u state
q n Cause Effects Conseq Task
resources
Planning
u c
Cause Effects one time
t
i i
•Procurement
Scheduling • Stocking
p o Failed
Cause one time • Lead times
Effects •Labor • Purchasing
m n state
•Skills decisions
e F Failed Yearly Sched. •Parts
n u
n
state
•Consumables
t c Quarterly Sched. •Outside
services
t
i Monthly Sched.
o •Task: tools, materials, •Tools
n safety procedures •Test equip
•Task: tools, materials,
safety procedures

87

History of CBM

87

The anatomy of CBM

Data Signal Decision

Acquisition Processing Making

91

17
 Data acquisition
HART (Highway Addressable Remote Transducer)
•A backward compatible enhancement to the 4-20mA
instrumentation installed in plants today. www.hartcomm.org
•Allows two-way communication with the smart
microprocessor based field devices that are now commonplace
• Carried on the same wires as, and not interrupting the 4-20
mA signal
• Provides access to the access to the wealth of information in
12 million HART devices.
•Process related variables are transmitted back as an IEEE
floating point values with engineering units and data quality www.mimosa.org
assessments.
• Supported by all of the major global instrument suppliers
MIMOSA (Machinery Information Management
Open System Alliance)
•Human-Machine Interfaces (HMI), Manufacturing Execution
Systems (MES), Plant Asset Management (PAM) systems,
Enterprise Asset Management (EAM) systems, Operational
Data Historian Systems (ODHS), and Condition Monitoring
(CM) systems.
•Common relational information system (CRIS)
www.osacbm.org OSACBM (Open System Architecture for
Condition Based Maintenance)
•UML
•AIDL
•IDL (CORBA, COM/DCOM, XML dotNET) 14

Notification logic – CBM triggers

• PI Alarm
• PI Performance Equation
• PI Advance Computing Engine

The anatomy of CBM

Data Signal Decision

Acquisition Processing Making

91

18
 Signal Processing

96

Signal Processing

Failure modes:
1. Shaft Rubs at bearings and seals due to oil whip,
2. coupling misaligned,
U.S. Patent No. 6460013 3. growth due to thermal effects,
4. lubrication loss,
5. oil contaminated,
6. blade erodes due to wet steam causing charge separation
and cavitation,
www.gaussbusters.com 7. charge separation and spark discharge due to dry steam at
inlet to turbine with partial admission,
8. shaft grounding lost,
9. intermittent ground fault due to torn copper leaf,
10. insulation shorted at bearings,
11. seals and couplings,
12. stator core lamination shorts,
13. diode fails in generator excitation,
14. excessive transients in pulse width modulated rotor and/or
stator electrical supply 96

Petri-nets

96

19
 Signal Processing

Many, ordinarily random signals, when represented in state space

using a branch of mathematics known as Chaos theory ,
display patterns, deviations from which may be tracked and related to
specific modes of failure.

97

Signal Processing

98

Signal processing … What next?

10

Where?

10

P-F Intvl
25/26 61/62 68 99

20
 The third sub-process of CBM
9Availability
9Cost
9Mission reliability
9Other KPI’s

Residual life
Decision Making!
estimate 56
days

100

CBM Fundamentals

CBM Program Criteria

103

Monitoring interval

Two BearingsTwo Bearings Noise

starts
Risk Functional Warning 2 wks
performance Very critical
Conditional Probability of Failure

OK Failed Inspection interval 1 week

Brg A
P-F = 2 Weeks

(MTBF = 3.5 years)

OK Warning 2 days
1/3.5
Noise Failed
Insp. interval 1
starts Not so critical
day
P-F = 2 Days
Brg B
1/7 (MTBF = 7 years)

Assertions:
1. The lower the Mean Time Between Failure (MTBF), the
more frequently you monitor?
2. The more critical, the more frequently you monitor?

103

21
 Initial inspection interval
Is CBM for the failure mode
in question applicable? (Is
there a clearly identifiable
condition indicator? Is the
warning time adequate?) NO

Is CBM for the failure mode

in question effective? (Is
there an economical CBM
task and interval that will YES
avoid or reduce, to a
tolerable level, the
consequences of the
failure?)

CBM not applicable or not

effective -> Descend to next
NO
task type in the RCM
algorithm.

YES

is the warning period of the

order of days, weeks, or
Initial inspection Interval =
months?
X/2

Days (Weeks, Months)

X Days (Weeks, Months)

How many days (weeks,
months)?

103

The Elusive P-F Curve

P-F curve and CBM optimization

106

The Elusive P-F

Interval
Inspection data
Inspection Ideal Real

*
Interval
Warning
* ?
Resistance to failure

Interval

P-F ?
Interval
Potential Failure
Failure P F
?
Working Age

106,108

22
 Modeling multiple risk factors
and working age

107

10

CBM Optimization

EXAKT

115

The Traditional CBM Model

Condition Indicator

Two Assumptions
1. Indicator reflects a failure mode degradation process.
2. The alarm limit is constant with age

Working Age

114

23
 Data and Risk
DATA PLOT
Data

Age

RISK PLOT
Risk

Age

115

At what level of risk to we want to

“maintain” an asset?

Cost/unit time Cost

Availability
Reliability
Reliability
Availability

High Risk High Risk

conservative laissez-faire
Lowest Highest Specified
cost availability reliabiltity

116

Typical Life of a
Component
E E E E E E
B B B B B B
F F F F F F

B = beginning of component life

EF = ending with failure

118

24
 No Scheduled Renewal Policy
CF

t1
CF
t2
CF
t3
CF
t4
CF
t5
CF

0 t6 6C F
Cost / hr =
t1 + t 2 + t3 + t 4 + t5 + t6

118

Preventivet Renewal Policy A

A

t1

t3

t4

t5

t6

5C R + 1C F
Cost / hr =
5t B + t6 119

Preventive renewal
t
policy B
B

t1

t3

t4

t5

t6

2C R + 4C F
Cost / hr =
2t B + t1 + t 2 + t3 + t6
120

25
 Cost calculation
Cost Per Unit Time for selected replacement costs
Policy NSM A: tA=2,500h B: tB=4,000h
Operational Time 20,000h 14,250h 19,000h

6 CF 5 CR + CF 2 CR + 4CF
Cost per unit
of working age 20,000h 14,250h 19,000h
Cost if CR=$2,000 $0.75/h $0.88/h $0.74/h
CF=$2,500
Cost if CR=$2,000 $1.5/h $1.06/h $1.26/h
CF=$5,000
Cost if CR=$2,000 $3.00/h $1.40/h $2.32/h
CF=$10,000

120

Q. Where to put the vertical

Policy
line?
POSSIBLE DECISION POLICIES:
no scheduled maintenance:
Policy B: 4000h
Policy A: 2500h
Policy A cost per
working hour:
Policy A t1
6CF
Policy B
t1 + t 2 + t3 + t4 + t5 + t6
t2

Policy B, cost per

Policy “At
t3
Failure
working hour:
Only” 5C R + 1C F
t4
5t A + t6
t5 Policy C, cost per
working hour:
2C R + 4C F
t6 tA = 2,500h tB = 4,000h

Sample 2t B + t1 + t 2 + t3 + t6
Lifetime 1 (F) 2 (F) 3 (F) 4 (F) 5 (F) 6 (F)
Duration 2750 3500 3000 4250 4750 1750 121

The Risk
Probability of
failure

Time

26
 Parameter Estimates

2.053−1
2.053  t 
h(t ) =  
11459.5  11459.5 

122

Where to put the vertical line? Or what is the best

time tp to do PM?
f(t) Area R(t) is the
probability that
Area F(t) is the the item will
probability of survive to time t.
failure at or F(t)
R(t)
before time t.
tp
ct = cR R (t p ) + cF (1 − R (t p )) tt = t p R(t p ) + t f (1 − R(t p ))

Ct cR R (t p ) + cF (1 − R (t p ))
= tp
tt t p R(t p ) + ∫0
t f (t )dt

The software finds the tp which minimizes the cost Ct

123-124

Does this
answer
question
t1
tp
posed earlier?
t2

t3

t4

t5

t6

125

27
 Proportional Hazards Modeling

Now, what would the optimal policy be if we have

the benefit of extra data - namely, condition data?

β −1
β t
h(t ) =   eγ 1 ( ppm Fe )+γ 2 ( ppm Pb )+γ 3 (ipsVib axial 2 rpm )+...
η η 

Parameters Condition data

values at time t

125

A Watchdog Agent
VA+OA+Other Event information
monitored data
Work orders

CMMS
CBM
data
Inspection

st on
Event data

Be ecisi
data

Decision
Expert Systems, neural nets, Case
Based Reasoning …
Model

126

From Data to Decision

127

28
 Example 1

128

Example 1 (slide 2)

128

Example 1 (slide 3)

129

29
Example 1 (slide 4)

130

Example 1 (slide 5)

131

Example 1 (slide 6)

131

30
 Example 2 (slide 1)

132

Example 2 (slide 2)

133

Building the Model - Graphical

Investigation

135

31
 More data investigations

ion
reg
his
in t
s i ng
a mis
Dat

135

Corrected Silicon

136

Oil changes

Fe B
ppm
C

A D Working age
Oil change interval

137

32
Missing oil change events

Missing ‘OC’ events?

138

Strange history

ry?
Histo
nge
Stra

139

Investigating the strangeness

No events to support this

jump in values

140

33
 The Definition of Failure
A major challenge in CBM optimization

Initially a failure was declared when wear metals were high.

This was like forcing the model to “chase its tail”.

Needed a physical definition of failure based on the

observable condition of the wheel motor at overhaul.

Based on the new definition (gear damage), the number of

histories ending in failure doubled.

Model “fit” improved dramatically.

140

The Wheelmotor
Optimal CBM Model

141

Had we replaced at 11384 hr…or 11653 hr…!!!!

Failed at WorkingAge = 11660 hr

Inspection at
WorkingAge = 11384 hr

Inspection at
WorkingAge = 11653 hr

142

34
 Profitability Impact of Optimized CBM

CR = 3
25% $1M

CR = 5
44% $1.7M

CR = 6
50% $2M

143

Sensitivity
Analysis

144

Example 3 – Complex items

Inspections_MA Events_MA EventsDescription_MA VarDescription_MA CovariatesOnEvent_MA
Ident Ident EventName VariableName Event
Date Date P MeasureUnit StartingDate
WorkingAge WorkingAge Comment WarnLimit1 EndingDate
Covariate1Name Event WarnLimit2 Covariate1Name
Covariate2Name Comment … Covariate2Name
… Comment …
Comment Comment

IdentToModel EventToModel VarToModel

ModelName ModelName ModelName

IdentName InputEventName InputVariableName
Date OutputEventName OutputVariableName
InputP VariableDataType
OutputP MeasureUnit
WarnLimit1
WarnLimit2
…

147

35
 Example 3 (2)

148-9

Example 3

150

The onion skins of CBM

CBM

CBM
inspection

.
. CBM
. overhaul

108, 268

36
11

The ABB Fault Simulator

Combining EXAKT with a

Vibration Expert system

152

CBM decisions with an Expert System

152

Cepstrum and
spectrum

153-5

37
 Demodulation

156-7

Expert rules

158

Decision making

160

38
 The ABB Fault Simulator

161-2

The ABB Fault Simulator

163

18

Managing strategy

Measuring and adapting KPIs

254

39
 Two improvement models

The Aladon RCM “House”

The Pyramid of Excellence

254

Managing Strategy

Visio
n
Strate
g y
Balance sheet
Maintenance Policy (market share, etc)
KPIs
•Proactive (PM, CBM, FF)
•No scheduled maintenance
Responsible
•Redesign custodianship

256

ABB
KPI Incident 1
Real-TPI

Function,
Quality Failure,
Cause,
Loss Effects,
Consequences IVARA
Malfunction
Machine
Process

EXP
SpotLight
CBR Improved
Incident 2 maintenance
Conditional probabilty of failure for 200

0.4
OSISoft
Function, PI
Failure,
0.3
policies
Cause, 0.2
Etc…
hour intervals

Total removals
Effects, 0.1 Potential failures

Consequences 0 1000
Functional failures
2000 3000 4000
Operating age since last shop visit (flight hours)

EXAKT
DLI
…

ExpertAlert
Incident “n”

RCKB

261

40