Introduction

Basic Maintenance Strategy of Engineering Assets)

Dr. Achmad Widodo

Mechanical Engineering Department

Diponegoro University

1
 Introduction to Condition Monitoring and Fault Diagnosis

What is Machines Condition Monitoring …?

 Basically, it is a maintenance tool

 Also, being applied in quality control, process control and process
monitoring

Diponegoro University 2 Mechanical Engineering Dept.

 Introduction to Condition Monitoring and Fault Diagnosis

ISO definition:
 A field of technical activity in which selected physical
parameters, associated with machinery operation, are
observed for the purpose of determining machinery integrity.

BS: 3811:1984
 The continuous or periodic measurement and interpretation
of data to indicate the condition of an item to determine the
need for maintenance.

Is it only based on vibration signal…?

Diponegoro University 3 Mechanical Engineering Dept.

 Introduction to Condition Monitoring and Fault Diagnosis

Instead of vibration signal, it also includes:

 Oil analysis (oil quality, contamination)

 Temperature
 Force
 Sound pressure (intensity)
 Wear particle monitoring and analysis
 Output of machine performance
 Product quality
 Visual inspection and others

Diponegoro University 4 Mechanical Engineering Dept.

 Introduction to Condition Monitoring and Fault Diagnosis

“Machinery Failure”
In context with the larger plant maintenance picture,
most machinery is required to operate within a close
set of limits
 Operating speed (not variable)
 Load condition
 Product quality standards

Cause of failure, i.e.:

Occasionally, machinery is required to operate outside
these limits for short times (electric generators).

Diponegoro University 5 Mechanical Engineering Dept.

 Introduction to Condition Monitoring and Fault Diagnosis

“Machinery Failure”
The main reason for employing a Machine Condition Monitoring
and Fault Diagnosis program is to find accurate, quantitative
information on the present condition of the machinery.

 Reasonable expectation of machine performance

 Will a machine stand a required overload?
 Should equipment be serviced now or later?
 What is the expected time to failure?
 What is the expected failure mode?

What is machinery failure…??

Diponegoro University 6 Mechanical Engineering Dept.

 Introduction to Condition Monitoring and Fault Diagnosis

“Machinery Failure”

Machinery failure is the inability of the machine to perform

its required function.
Machine failure is specific.

 Belt conveyor drive-end bearing

Failure = seized bearing (belt stop)
 Computer disk-drive
Failure = slow response, noisy
 Rotating machines
Failure = excessive vibration

Diponegoro University 7 Mechanical Engineering Dept.

 Introduction to Condition Monitoring and Fault Diagnosis

“Machinery Failure”
Other considerations which may dictate the machine
performance:
 Economic the obsolescence-run to failure, premature
replacement
 Safety  minimizing risk of failure: train, airplane, ship,
automobile, etc.

Diponegoro University 8 Mechanical Engineering Dept.

 Introduction to Condition Monitoring and Fault Diagnosis

“Machinery Failure”

Causes of failures:

 Design and manufacture deficiencies

 Material deficiencies
 Processing deficiencies
 Improper assembly practices
 Improper service conditions
 Inappropriate maintenance
 Excessive demands

Diponegoro University 9 Mechanical Engineering Dept.

 Introduction to Condition Monitoring and Fault Diagnosis

“Types of Failure”

 Catastrophic: sudden and complete

 Incipient: partial, usually gradual

In most cases there is some advanced warning of

the onset of failure.
All failures pass through an incipient phase, even
if they do so quickly.

Diponegoro University 10 Mechanical Engineering Dept.

 Introduction to Condition Monitoring and Fault Diagnosis

“The goals”

 Detect onset of equipment deterioration

 Diagnose the condition
 Trend its progression over time
 Prognosis (predict) when ultimate failure will occur
 Allow time for maintenance planning

This excludes failures caused by unforeseen and

uncontrollable outside forces - earthquakes, etc.

Diponegoro University 11 Mechanical Engineering Dept.

 Introduction to Condition Monitoring and Fault Diagnosis

“Frequency of Failure”

Bath-Tube Curve, for individual or population of machines

Diponegoro University 12 Mechanical Engineering Dept.

 Introduction to Condition Monitoring and Fault Diagnosis

“Wear In Failure”
Typically high frequency of failures early in the expected
life of a machine due to:

 Design errors
 Manufacturing defects
 Assembly mistakes
 Installation problems

Diponegoro University 13 Mechanical Engineering Dept.

 Introduction to Condition Monitoring and Fault Diagnosis

“Random Failure”

 Relatively low failure rate when operating within

design specifications

Diponegoro University 14 Mechanical Engineering Dept.

 Introduction to Condition Monitoring and Fault Diagnosis

“Wear Out Failure”

Occur towards the end of a machine’s design life.

Gradually increasing failure rate at the expected
end of a machines useful life - primarily due to:

 Fatigue
 Wear mechanisms
 Corrosion
 Obsolescence

Diponegoro University 15 Mechanical Engineering Dept.

 Introduction to Condition Monitoring and Fault Diagnosis

“Wear Out Failure”

The slope of the wear-out part of the bath-tub curve is

machine and operational history dependent.
 If the machine design is such that the operational life ends abruptly
(due to fatigue for example) or the machine is under designed to
meet the load expected or the machine has endured a severe
operational life (experienced numerous over-loads) the slope of the
curve in the wear out section will increase sharply with time. If the
machinery is over designed or experiences a relatively light loading
history the slope of this part of the bath-tub-curve will increase
only gradually with time.

This curve shape is generally true for individual

machines and populations of machines of the same type.

Diponegoro University 16 Mechanical Engineering Dept.

 Introduction to Condition Monitoring and Fault Diagnosis

“Frequency of Failure”

Diponegoro University 17 Mechanical Engineering Dept.

 Basic Maintenance Strategy
“Run to Failure Maintenance”
 Maintenance performed only when machinery has failed.
 Example: Wear out of machine tool

Diponegoro University 18 Mechanical Engineering Dept.

 Basic Maintenance Strategy
“Scheduled Maintenance”

 Specific maintenance tasks performed at set time intervals

(or duty cycles).
 Significant margin between machine capacity and actual duty
maintained.
 Example: Oil change on your car

Diponegoro University 19 Mechanical Engineering Dept.

 Basic Maintenance Strategy
“Condition Based Maintenance”

 Based on predictive, condition, pro-active, reliability

centered, etc.
 actual condition of the machinery is assessed
 data used to optimally schedule maintenance
 maximum production and avoidance of catastrophic failures
is achieved
 Example: Tire chance on your car

Diponegoro University 20 Mechanical Engineering Dept.

 Basic Maintenance Strategy
“Condition Based Maintenance”

 Note: margin between duty and capacity is never allowed to

reach zero – breakdown avoidance.
 Results: longer time between maintenance tasks than for
scheduled maintenance.

Diponegoro University 21 Mechanical Engineering Dept.

 Basic Maintenance Strategy

 Advantages and disadvantages do exist.

 Situations exist where one or the other would be
appropriate.
 The maintenance engineer must decide and justify action.
 Combinations of strategies may also be required.
 Examples:
 increased frequency of monitoring as the age of a
machine increases.
 maximum time between overhauls with monitoring
looking for random failures

Diponegoro University 22 Mechanical Engineering Dept.

 Basic Maintenance Strategy

Open Source Architecture-CBM

(OSA-CBM)
Physical Model
Data
acquisition
S1
Data Health Automatic
Complex Prognostic Decision
Manipulation Condition Assessment
system S2
Reasoning Human
Monitor
- Preprocessing - Component - Feature based
prognosis
Computer
- Feature specific feature
extraction
- Thresholds
extraction - Model based
- Data fusion
- Classifier
Interface
- Fuzzy logic prognosis
- Signal - Anomaly and - Response
characterization diagnosis resoners
S3 generator

7
1 2 3 4 5 Mission plans

Diponegoro University 23 Mechanical Engineering Dept.

 Basic Maintenance Strategy

Diponegoro University 24 Mechanical Engineering Dept.

 Basic Maintenance Strategy

Evolution of applied maintenance system

<1950 1950-1975 >1975 2000-
Manpower (simple) Mechanization Automation (more Globalization (crossing
(complex) complex) boundaries)
“Fix it when it breaks” “I operate—you fix” RAM (Safety, Quality, Optimal concept + Outsourcing
(Availability, Environment) CBM, CM, & ICT
Longevity, Cost) DOM, Mutli-skilling,
PM,WO-mgmt. MMIS Asset mgmt
Maintenance is “A Maintenance is “A Maintenance is “(maybe) Maintenance is “External and
production task” task of the Not an isolated function internal partnerships”
maintenance dept.” Integration efforts” Maintenance meets production
“Necessary evil” “Technical matter” “Profit contributor” “Partnership”

20
Maintenance cost ($/HP/Year)

15

10

0
RTF PM PdM PCM
Run to Preventive Predictive Proactive Condition
Failure Maintenance Maintenance Monitoring

Diponegoro University 25 Mechanical Engineering Dept.

 Basic Maintenance Strategy
“Factors Which Influence Maintenance Strategy”

 Classification of machine
- Critical to production… ?
- High cost of replacement…?
- Long lead time fro replacement…?
 Manufacturers recommendations
 Failure data (history), MTTF, MTBF, failure modes
 Redundancy
 Safety (plant personnel, community, environment)
 Parts cost/availability
 Costs (personnel, administrative, equipment)
 Running costs

Diponegoro University 26 Mechanical Engineering Dept.

 Basic Maintenance Strategy
“Factors Which Influence Maintenance Strategy”

MTBF 
 (downtime  uptime)
Number of failures

The MTBF is typically part of a model that assumes the failed system is immediately
repaired (zero elapsed time), as a part of a renewal process.
This is in contrast to the mean time to failure (MTTF), which measures average time
between failure with the modeling assumption that the failed system is not repaired.

Diponegoro University 27 Mechanical Engineering Dept.

 Basic Maintenance Strategy
“Factors Which Influence Maintenance Strategy”
Considerations for breakdown maintenance:

 If equipment is redundant
 Low cost spares available
 Interruptible process, stockpiled product
 Safe failure modes
 Long MTTF/MTBF
 Low cost secondary damage
 Quick repair or replacement (low cost of interruption to
production)

Diponegoro University 28 Mechanical Engineering Dept.

 Basic Maintenance Strategy
“Factors Which Influence Maintenance Strategy”
Considerations for predictive maintenance:

 Statistical failure rate available

 Narrow failure distribution (predictable MTBF)
 Maintenance restores full integrity
 Single failure mode (known)
 Low cost of regular overhaul/replacement
 Unexpected interruptions to production
 Expensive (scheduled interruptions not so bad)
 Low cost spares available
 Reduced number of breakdowns required
 Costly secondary damage from failure

Diponegoro University 29 Mechanical Engineering Dept.

 Basic Maintenance Strategy
“Factors Which Influence Maintenance Strategy”
Considerations for condition based maintenance:
 Expensive/critical machinery
 Long lead time for replacement (no spares)
 Uninterruptible process (both regular and unexpected) –
costly
 Large/complex machinery
 Overhaul expensive/needs highly trained people
 Reduced numbers of highly skilled maintenance people
 Costs of monitoring program acceptable.
 Safety is a priority (failures dangerous)
 Remote, mobile equipment
 Failure not indicated by operation degeneration
 Costly secondary damage.
Diponegoro University 30 Mechanical Engineering Dept.
 Machine Condition Monitoring and Fault Diagnosis

“Potential Advantages”
 Increased machine availability and reliability
 Improved operating efficiency
 Improved risk management (less down time)
 Reduced maintenance costs (better planning)
 Reduced spare parts inventories
 Improved safety
 Improved knowledge of machine condition (safe overloading
of machine possible)
 Extended operational life of machine
 Improved customer relations (less planned /unplanned
downtime)
 elimination of chronic failures (root cause analysis and
redesign)
Diponegoro University 31 Mechanical Engineering Dept.
 Machine Condition Monitoring and Fault Diagnosis

“Potential Disadvantages”

 Monitoring equipment costs (high)

 Operational costs (running the program)
 Skilled personnel needed
 Needs strong management commitment
 Long run-in time to collect machine histories and set trends
 Reduced costs are harder to sell as direct benefits to
management than increased profits

Diponegoro University 32 Mechanical Engineering Dept.

 Machine Condition Monitoring and Fault Diagnosis

“Philosophy”
Get useful information on the condition of equipment
to the people who need it.
 Operators, maintenance, managers, etc.
 These groups need different information at different time

This means:
 Collect useful data
 Change data into information in a form required by and
useful to others
 Timely reporting

Diponegoro University 33 Mechanical Engineering Dept.

Machine Condition Monitoring and Fault Diagnosis

“Philosophy”

Types of data collected:

 Vibration severity, frequency analysis, temperature, oil
analysis, etc.
Types of information gleaned:
 existing condition
 Trends
 expected time to failure at a given load
 type of fault existing or developing
 type of fault which caused failure

Diponegoro University 34 Mechanical Engineering Dept.

 Machine Condition Monitoring and Fault Diagnosis

“Tasks”
 Detection
• Data gathering
• Comparison to standards
• Comparison to limits set in-plant for specific equipment
• Trending over time

 Diagnosis
• Recognizing the type of fault developing (different fault
types may be more or less serious and require different
action)
• Severity of fault

Diponegoro University 35 Mechanical Engineering Dept.

 Machine Condition Monitoring and Fault Diagnosis

“Tasks”
 Prognosis
• Expected time to failure
• Trending
• Forecasting
• Maintenance planning/timing

 Post Mortem
• Root cause failure analysis
• Research, laboratory / field tests
• Modeling of system and analysis

Diponegoro University 36 Mechanical Engineering Dept.

 Machine Condition Monitoring and Fault Diagnosis

“Tasks”
 Prescription (activity dictated by information collected)
• May be applied at any stage
• Alter operating conditions
• Alter monitoring strategy (frequency, type)
• Redesign process or equipment

Diponegoro University 37 Mechanical Engineering Dept.

 Machine Condition Monitoring and Fault Diagnosis

“Strategies”

 How much data to collect?

 How much time to spend at data analysis?
• These things dictate the MCM&FD strategy (cost will
always be a factor)
 Consider:
• Equipment class, size, importance within process,
replacement cost and availability
• Safety
• Different pieces of equipment or processes may require
different monitoring strategies.

Diponegoro University 38 Mechanical Engineering Dept.

 Machine Condition Monitoring and Fault Diagnosis

“Strategies”

No Monitoring
 Inexpensive, non-critical equipment
 In stock equipment (or readily accessible)
 Low load equipment
 Low failure rate known
 Failure modes well understood

Diponegoro University 39 Mechanical Engineering Dept.

 Machine Condition Monitoring and Fault Diagnosis

“Strategies”

Periodic Monitoring
 Non-critical equipment
 Failure modes known
 Historically dependable equipment
 Trending and severity levels checks only
 Problems trigger more rigorous investigations

Diponegoro University 40 Mechanical Engineering Dept.

 Machine Condition Monitoring and Fault Diagnosis

“Strategies”

Continuous Monitoring

 Permanently installed monitoring system samples and

analyses data automatically
 Critical equipment (expensive to replace with downtime (loss
of production) being expensive)
 Changes in condition trigger more detailed investigation or
possibly automatic shutdown

Diponegoro University 41 Mechanical Engineering Dept.

 Machine Condition Monitoring and Fault Diagnosis

“What to Measure”

 Vibration levels (displacement, velocity, acceleration).

 Oil analysis (lubricating quality contamination)
 Wear particle monitoring and analysis (number, size, shape,
composition)
 Force measurements
 Sound level
 Odour
 Temperature
 Output quantity
 Product quality
 Visual inspection, etc.

Diponegoro University 42 Mechanical Engineering Dept.

Machine Condition Monitoring and Fault Diagnosis

“Framework”

General method

Sensor Decision Failure

Residual generation Residual Information collection Decision making
outputs statistic decision

Model-based Data-driven

Sensor readings Sensor readings

Residual or Knowledge and

Model Inferencing
innovations training

Decision logic Decision logic

Diponegoro University 43 Mechanical Engineering Dept.

 Machine Condition Monitoring and Fault Diagnosis

“Framework”
Online fault diagnosis system
Refined diagnosis

Time Monitoring
Sensor outputs Data-driven
waveform, FFT Feature information
vectors Identify Yes
Digital signal Yes
faults/failures
processing Feature Fault classification
Machines vectors and  (t ) Feature patterns for faults More info
No
Decision:
statistic Decision fusion could use: needed? Good?
System  information Fuzzy, NN, SVM, etc.
identification No
Mathematic
model Physical parameters Request
Model-based estimation
Physic of maintenace
Sensor fusion Feature fusion
failure; System


dynamic
parameters Feature extraction

Diponegoro University 44 Mechanical Engineering Dept.

 Machine Condition Monitoring and Fault Diagnosis

“Problem during Implementation”

Problem encountered:
 No associates training on maintenance know-how for top persons which results
in lack of management support
 Lack of operator’s comprehension
 Lack of resources
 Lack of involvement of production associates
 Lack of long-term vision; quick return expected by management
 Lack of sustained momentum

Complicated problem as follows:

 A symptom can be caused by different fault conditions
 Some faults are not easy to recognize in the machine, due to the
background noise
 There are many components with machinery.
 There is a high level of interaction between these components

Diponegoro University 45 Mechanical Engineering Dept.