SoMMME

Department of Mechanical Engineering

PME 3132: Maintenance Engineering and

Management

Introduction to Condition Monitoring

James K. Kimotho
 SoMMME
Department of Mechanical Engineering

Content

Motivation for Condition Monitoring

Introduction
Diagnostic Methods
Prognostic Methods

2 PME 3132/ Condition Monitoring/ J.K. Kimotho / 20.06.2019

 SoMMME
Department of Mechanical Engineering

Motivation

Advances in information technology has led to the development of low

cost microprocessors and multi-functional data acquisition systems
with direct sensor connectivity.
Advances in sensor technology, automated controls as well as data
telemetry have led to innovative and low cost sensors for monitoring
operation characteristics such as
– electric current, pressure, flow, temperature as well as for condition
monitoring such as tribology, thermography, vibration, acoustics and force
Technologies in remote condition monitoring have made it possible to
monitor systems in locations that are not easily accessible.
– A state of the art real-time monitoring system referred to as Supervisory
Control and Data Acquisition (SCADA) has been extensively employed in
real-time performance and remote condition monitoring of complex
industrial systems
CM aims at avoiding inopportune maintenance spending in addition to
maximizing component and system life

3 PME 3132/ Condition Monitoring/ J.K. Kimotho / 20.06.2019

 SoMMME
Department of Mechanical Engineering

Motivation
Question: Is it possible to anticipate failures during operation?
Answer: Yes through condition monitoring
What is condition monitoring?

[KAt]

5
0.2 Threshold 0.2 Threshold
Vibration / g

Vibr. Ampl. / g

Vibr. Ampl. / g
0
0.1 0.1

-5
0 0
0 0.5
0 100 200 0 100 200
Time / s Frequency / Hz
Frequency / Hz

What are the benefits?

– Increased reliability, safety and availability
– Improved product quality through performance monitoring
4 PME 3132/ Condition Monitoring/ J.K. Kimotho / 20.06.2019
 SoMMME
Department of Mechanical Engineering

Introduction
Advancement of condition monitoring
– Diagnostics: detection, isolation and identification of faults

0.2 Threshold

Vibr. Ampl. / g
0.1

0
0 100 200
[KAt] Frequency / Hz

– Prognostics: estimation of current health state and predicting failure

• Estimating the remaining useful lifetime (𝑅𝑅𝑅𝑅𝑅𝑅)

𝑅𝑅𝑅𝑅𝑅𝑅 Operation Time

Now Time at failure

– Prognostics and health management (PHM) includes a decision module

into diagnostics and prognostics

5 PME 3132/ Condition Monitoring/ J.K. Kimotho / 20.06.2019

 SoMMME
Department of Mechanical Engineering

Introduction

Sensed data contain information about the condition of

components, systems, and processes
– A lot of data are being collected all the time
– Equipment data, process parameters, operating conditions …
We want to
– discover the underlying relationships in data
– exploit these relationships to make predictions or decisions about
new data

6 PME 3132/ Condition Monitoring/ J.K. Kimotho / 20.06.2019

 SoMMME
Department of Mechanical Engineering

Diagnostic Methods
Diagnosis is an assessment of the current health state of a
system based on observed features from condition monitoring
data.
– It involves identifying anomalous behavior in a system, locating
the component with anomalous behavior and identifying the type
of anomaly in that component.
– In some instances, diagnosis can imply the isolation of a faulty
component, a failure mode, or a failure condition
Two main categories:
– Data driven diagnostics
– Model based diagnostics

7 PME 3132/ Condition Monitoring/ J.K. Kimotho / 20.06.2019

 SoMMME
Department of Mechanical Engineering

Diagnostic Methods
Data-driven Diagnostics

Two sub-categories
– Expert knowledge with signal processing
– Machine learning algorithms
Advanced signal processing coupled with expert knowledge of
the system is used to identify possible faults.
Graphical tools such as power spectrum graph, cepstrum
graph, spectrogram, etc, are employed.
In rotating machinery, fault characteristic frequencies such as
bearing frequency characteristics, gear mesh frequencies,
shaft frequencies and their harmonics, etc, have been
extensively used in fault diagnosis

8 PME 3132/ Condition Monitoring/ J.K. Kimotho / 20.06.2019

 SoMMME
Department of Mechanical Engineering

Diagnostic Methods
Application Example - Rolling Element Bearing Faults

Rolling element bearings account for the largest percentage of

failures in rotating machinery

9 PME 3132/ Condition Monitoring/ J.K. Kimotho / 20.06.2019

 SoMMME
Department of Mechanical Engineering

Diagnostic Methods

Bearing characteristic frequencies:

10 PME 3132/ Condition Monitoring/ J.K. Kimotho / 20.06.2019

 SoMMME
Department of Mechanical Engineering

Diagnostic Methods
Data-driven Diagnostics – Machine Learning
Converts sensor data into useful Advantages
information – No expert knowledge of system
Application of machine learning – Easily adaptable to different
algorithms systems
– Classification Disadvantages
– Clustering – Requires huge amount of data for
- Fault class
- Health States training
– Susceptible to overfitting
Training
Target

Machine
Training Feature
Learning Model
Data Extraction
Algorithm

Online application

New Feature
Model Output
Data Extraction

11 PME 3132/ Condition Monitoring/ J.K. Kimotho / 20.06.2019

 SoMMME
Department of Mechanical Engineering

Diagnostic Methods
Model based Diagnostics
Detailed mathematical models Advantages
– Physics of failure – High accuracy
Empirical based models – Can be implemented at
development stage
Disturbances Faults Disadvantages
𝑑𝑑(𝑡𝑡) 𝑓𝑓(𝑡𝑡) – Requires expert knowledge of
system
Inputs Actual Output
– Computationally intensive
𝑢𝑢(𝑡𝑡) 𝑦𝑦(𝑡𝑡)
– System specific
system

Residual
+ 𝑟𝑟(𝑡𝑡)
𝐺𝐺[𝑢𝑢 ∙ , 𝜃𝜃]
Diagnostics
Model -
Expected Output
𝑦𝑦�(𝑡𝑡)

12 PME 3132/ Condition Monitoring/ J.K. Kimotho / 20.06.2019

 SoMMME
Department of Mechanical Engineering

Prognostics: Term Definitions

Methods used to predict:

– Remaining Useful Life (RUL): the amount of time, in terms of
operating hours, cycles, or other measures, the component will
continue to meet its design specification.
– Time of Failure (TOF): the time a component is expected to fail
(no longer meet its design specifications).
– Probability of Failure (POF): the failure probability distribution of
the component.
– Trending: Trend or linearly project/regress a current measurement
until it reaches a predefined threshold

13 PME 3132/ Condition Monitoring/ J.K. Kimotho / 20.06.2019

 SoMMME
Department of Mechanical Engineering

Prognostic Term Definitions

Remaining Useful Life (RUL) – The amount of time a

component can be expected to continue operating within its
stated specifications.
– Dependent on future operating conditions
• Input commands/conditions
• Environment
• Loads

14 PME 3132/ Condition Monitoring/ J.K. Kimotho / 20.06.2019

 SoMMME
Department of Mechanical Engineering

Prognostic Methods

Categories of prognostic methods

Model Reliability
based based

Data driven Hybrid

based

15 PME 3132/ Condition Monitoring/ J.K. Kimotho / 20.06.2019

 SoMMME
Department of Mechanical Engineering

Prognostic Methods
Model based prognostics
Detailed mathematical models Advantages
– Physics of failure – High accuracy
Empirical based models – Can be implemented at
development stage
Disturbances Faults Disadvantages
𝑑𝑑(𝑡𝑡) 𝑓𝑓(𝑡𝑡) – Requires expert knowledge of
system
Inputs Actual Output
– Computationally intensive
𝑢𝑢(𝑡𝑡) 𝑦𝑦(𝑡𝑡)
– System specific
system

Residual
+ 𝑟𝑟(𝑡𝑡) 𝑅𝑅𝑅𝑅𝑅𝑅
𝐺𝐺[𝑢𝑢 ∙ , 𝜃𝜃] Damage Damage
Diagnostics
Model - Estimation Propagation
Expected Output
𝑦𝑦�(𝑡𝑡)

16 PME 3132/ Condition Monitoring/ J.K. Kimotho / 20.06.2019

 SoMMME
Department of Mechanical Engineering

Prognostic Methods
Model based prognostics

Description of a system’s underlying physics using suitable

representation
Some examples:
– Model derived from “First Principles”
• Encapsulate fundamental laws of physics
– PDEs
– Euler-Lagrange Equations
– Empirical model chosen based on an understanding of the
dynamics of a system
• Lumped Parameter Model
• Classical 1st (or higher) order response curves
– Mappings of stressors onto damage accumulation
• Finite Element Model
• High-fidelity Simulation Model
• Something in the model correlates to the failure mode(s) of interest
17 PME 3132/ Condition Monitoring/ J.K. Kimotho / 20.06.2019
 SoMMME
Department of Mechanical Engineering

Prognostic Methods
Data driven based prognostics
Converts sensor data into useful Advantages
information – No expert knowledge of system
Application of machine learning – Easily adaptable to different
algorithms systems
– Classification Disadvantages
– Regression - RUL – Requires huge amount of data for
- Health Index training
- Health States
– Susceptible to overfitting
Training
Target

Machine
Training Feature
Learning Model
Data Extraction
Algorithm

Online application

New Feature
Model Output
Data Extraction

18 PME 3132/ Condition Monitoring/ J.K. Kimotho / 20.06.2019

 SoMMME
Department of Mechanical Engineering

Prognostic Methods
Reliability based prognostics
Utilizes distributions of historical failure times of similar systems
Usually for low risk systems with no sensor network
Commonly employed distributions:
– Weibull
– Exponential
– Lognormal
– Normal
Extensively applied in prognostics of electronic systems
Advantages
– No condition monitoring data is required
– Easy to implement
Disadvantages
– Low accuracy
– No information on incipient faults
– Operating conditions are not taken into account
– Difficult to implement at component level
19 PME 3132/ Condition Monitoring/ J.K. Kimotho / 20.06.2019
 SoMMME
Department of Mechanical Engineering

Prognostic Methods
Reliability Based Prognostics

Estimate TTF

-3
10
1 6

Failure PDF
0.8
2

0
0.6 0 100 200 300 400 500
Time (days
Reliability

0.4

0.2

0
0 50 100 150 200 250 300 350 400 450 500
Time (days

20 PME 3132/ Condition Monitoring/ J.K. Kimotho / 20.06.2019

 SoMMME
Department of Mechanical Engineering

Tutorial
Tutorial 3.1

Data sets Tutorial_3a, Tutorial_3b and Tutorial_3c contain

vibration signals from normal, outer ring fault and inner ring
fault respectively. The bearing geometry is given in the table
below.
– Convert the time signal into frequency spectrum.
– Identify the bearing characteristic frequencies.
– Is the method effective in detecting faults?

The data is structured with the following fields:

– Sampling rate “sr”
– Vibration signal in g, ‘gs’
– Shaft speed in Hz, “rate”
21 PME 3132/ Condition Monitoring/ J.K. Kimotho / 20.06.2019
 SoMMME
Department of Mechanical Engineering

Tutorial
Tutorial 3.2

The Table below shows the failure times of similar

electromechanical actuators. Assuming the TTF has Weibull
distribution,
– determine the distribution parameters,
– If the current operating time for an actuator is 50 days, determine
the reliability of the component.
– Estimate the remaining useful lifetime of the component and its
distribution.

22 PME 3132/ Condition Monitoring/ J.K. Kimotho / 20.06.2019

 SoMMME
Department of Mechanical Engineering

References

Vachtsevanos, G.; Intelligent Fault Diagnosis and Prognosis for

Engineering Systems, John Wiley and Sons, 2006.
Kimotho, J.; Development and Performance Evaluation of
Prognostic Approaches for Technical Systems, University of
Paderborn, 2016.
Vachtsevanos, G. et al.; Tutorial: Introduction to Prognostics,
PHM Society Conference, 2015.

23 PME 3132/ Condition Monitoring/ J.K. Kimotho / 20.06.2019

 SoMMME
Department of Mechanical Engineering

Thank You
for
Your Kind Attention!
Dr.-Ing. James Kuria Kimotho Department of Mechanical Engineering
[email protected] JKUAT
P.O. Box 62000-00200
Tel: Nairobi