SlideShare a Scribd company logo
PREDICTING FAILURES 
Ir. Gan Chun Chet 
August 2007
The Concept and Practice 
• In the Meeting Room 
Figures Are Showing 
Better Outlook 
• The Ability to Know Bad 
Results Before It 
Happens, Would Be 
Good to Know Before It 
Occurs 
• Hence, A Scientific 
Method to Measure Past 
Records to Achieve the 
Required Future Targets 
Emerge 
I have 
done 
enough. 
Engineer 
Everything 
Seems to Be 
OK. 
Plant 
Manager 
Profit 
is 
Rising 
I’m Impress 
Managing 
Director 
PREDICTING 
FAILURES
Predicting Failures 
• What is Predicting? 
• The ability to know something before it 
happens 
• What are Failures? 
• Something that does not pass the required 
results
Predicting Failures 
• Some items that are highlighted in this talk 
- Plant Noise 
- Pumps and Compressors Vibration 
- Machine Setup Time 
- Redundancy 
- Partial/Full Stroke Testing 
- Review/Check Activity
Predicting Failures 
• Introduction 
• Passes, Failures and Predictions 
• Some Examples To Serve as A Guide 
• Some Counter Measures to Overcome 
• Conclusion
Introduction 
• Practices in General Industry Engineering 
Project and Maintenance Services 
- Engineering Best Practices 
• Oil and Gas Consultancy and Contracting 
Services 
- Requires High Safety Standards 
- Requires High Reliable Products
Introduction 
• Some of the Issues Addressed in General Industry 
(Manufacturing and Processing) 
- Hearing Losses 
- Pumps and Compressors Concerns of A Breakdown 
- Long Machine Setup Time Which Contributed to 
Uneasiness Among Production Personnel (Production 
Planning, Production Morale, Machine Change Over, 
etc.) 
• Some of the Practices from the Oil and Gas Industry that 
worth while knowing 
- Redundancy 
- Partial/Full Stroke Testing 
- Review Check Activities
General Industry - Engineering 
Project and Maintenance Services 
• Improve Plant Conditions 
- Safe Working Areas 
Inhabitants in the plant are able to work in a healthy and 
conducive environment 
• Reduce Breakdowns 
- Continuous Batch Operation 
The production purpose is to ensure continuous supply 
to the market due to demand (to operate profitably)
Oil and Gas - Consultancy and 
Contracting Services 
• Query Services 
- Latest Development 
Client would like to know whether the latest product is 
purchased 
• Design and Engineering Services 
- Specify the Right Requirements 
- Integrating Properly 
Engineered correctly to enable the platform or plant 
operating within the design production output, a proper 
detection system, the best way to fight a fire, equipment 
are maintain to it best performance, etc.
What Causes Failures? 
• Causes of Failures 
- Wear and Tear (Bath Tub Theory) 
- Vibration 
- Exhaustive Operation 
• Are Failures Detectable? 
- Can it be Spotted and Rectified? 
- Are there Methods to Detect Failures? 
(i.e. To know before something happen)
Passes, Failures and Predictions 
• 100% Passes 
- Is This Achievable 
(Take note of Process Variance) 
• 0% Failures 
- Is this achievable 
(Perfection? – Good Engineering Design) 
• T+1 Prediction 
- Based on Facts and Past Data
Passes 
• National or University Examination Results 
- Number of Students with Distinction 
Is increasing the passing rate an indication of better 
quality students? 
University in the search of glooming student that are able 
to meet today’s challenges 
• Productivity 
- Operation Output 
The measurement of productivity is to achieve better 
result each year until it reaches its best performance. 
Business/Management will begin to look for other 
performance parameters
Passes 
• Upper Range Limit 
- Highest Acceptable Specification 
Value 
• Lower Range Limit 
- Lowest Acceptable Specification 
Value 
120 
100 
80 
60 
• Is it possible to increase the 
40 
limit, hence more will pass 
• Take note of design quality 
20 
• Take note of process 
(machine) limit 0 
LRL URL 
0 5 10 15
Failures 
• Bath Tube Theory 
- Failures Occur Beginning of Design 
Life. 
- Sporadic Failures During Normal Life 
Stage. 
- Failures Begins to Occur Towards 
the End of Design Life. 
• Is This Actually 
Happening in Real 
Operation? 
7 
6 
5 
4 
3 
2 
1 
0 
0 10 20 
Time 
Failure Rate 
Infant 
Mortality 
Stage 
Wear-out 
Stage 
Normal-life 
Stage
Failures 
• Off-Spec Rejects 
- Not Meeting 
Specification 
- A Known Process 
Variation 
• Sporadic Peaks 
- Sudden 
Occurrences 
- Unpredictable 
150 
100 
50 
0 
300 
200 
100 
0 
1 2 3 4 5 6 7 8 9 1
Failures 
• Shutdown/Detection System Fail To 
Operate 
- Not Able to Detect or Shutdown 
• Valves Unable to Function 
- Stuck 
- Cannot Close
Failures 
• Failure Rate 
= Number of Failures X 100 
Total Number of Products Tested 
(p776, Operations Management – 
Nigel Slack et al, Pitman Publishing) 
• An equation to know the number of actual failure coming out of the 
production line 
• Measurement of the percentage of product failure
Failures 
• Mean Time Between Failure (MTBF) 
= Operating Hours 
Number of Failures 
(p776) 
The average time between failures, measured in actual operation. 
The number of breakdowns over the operation duration. 
(To Determine Time Before a Failure, Average of Past Record)
Prediction 
• Threshold Limit 
- Above a Set Horizontal Line Will 
Trigger a Warning. 
• Review of past data shows that a limit 
has to be set. Anything above this set 
limit must need to be attended. 
• E.g. Noise (Hearing) Level 
8 
7 
6 
5 
4 
3 
2 
1 
0 
Threshold 
Limit 
Warning 
0 5 10 15
Prediction 
• Analysis 
- Frequency Spectrum 
• An analysis of the pump or 
compressor vibration behaviour. 
Because the pump and compressor 
rotate at fixed angular rotation, the 
frequency of the pump or compressor 
can be calculated. 
The spectrum shows the frequency of 
the pump or compressor, etc. 
Frequency 
Amplitude 
f 2f 3f
Prediction 
• Trend Line 
- A Straight Line Can Be Drawn on The Collected Data. 
- The Straight Line Can be Extended to Know the Next 
Value. 
E.g. A Linear Relationship Between X and Y and be 
Concluded, Case of reduction in setup time. 
Line of Regression 
10 
5 
0 
0 5 10
Prediction 
• Moving Average 
- Average of Past Data and Moving 
Forward 
• The moving average line of 2 past data 
is the average of 2 past data. 
• The moving average line of 2 past data 
is more reactive than 5 past data. 
• Moving average is the record of past 
90 
80 
70 
60 
50 
40 
30 
historical data. With this type of trend 
line, the future up or down trend can 
20 
be know, provided all factors remain 
unchange. 
10 
(E.g. Monitoring Setup time) 0 
5 past data 
2 past data 
0 5 10 15
Real Examples 
• Threshold Shift (Noise Study) – (Factories 
and Machinery, Noise Exposure) 
Regulations, 1989 
(A Real Noise Reduction Project) 
• Noise Level Collected and Recorded at 500, 1000, 2000, 
3000 & 4000Hz 
• Sampled of Production Workers at Floor Level
Real Examples 
• Less than 10 years 
service 
• 2 Clusters 
• Results 
- Below 20 dB 
(Cluster 1) 
- Below 30 dB 
(Cluster 2)
Real Examples 
• More than 10 years 
service 
• 1 Cluster 
• Results 
- Below 30 dB
Counter Measures 
• Hearing Conservation Program 
- Engineering Control 
(To Install Enclosure At Rotating Equipment, To Reduce 
Device Noise e.g. Control Valve & Relief Valve, To 
Prevent Noise From Traveling Into The Control Room) 
Enclosure 
Machine 
Limit noise out of the 
enclosure 
Control Room 
Limit noise to the control 
room
Counter Measures 
• - Administrative Control 
(To Move Workers In And Out Of Noisy Area (take note 
of threshold shift, very dangerous, so act mentioned to 
limit daily exposure to noise by control of work schedule, 
very difficult to practice), Constant Monitoring, Do Not 
Allow People To Enter Noisy Areas, Sign Board of High 
Noise Areas, etc.) 
HIGH NOISE 
AREA 
- WEAR EAR 
MUFFLE
Counter Measures 
• Required by law that the Permissible Exposure Limit is 
- 85 dB(A) for 16 hours duration per day, 
- 90 dB(A) for 8 hours duration per day, etc per schedule 
- Nobody shall be exposured to noise level exceeding 
115 dB (A) at any time 
Factories And Machinery (Noise 
Exposure) Regulations, 1989 
• Employees need to check for a baseline reading 
(required by law) 
• Take note of Threshold Shift! DO NOT play with this! 
• Reduce Noise Level (Engineering Measures) 
• Consult Noise Consultant or Ear Specialist!
Counter Measures 
• Some Examples (Also Consult Doctors or 
Specialist! Hearing Problem Must not be 
Taken Lightly) 
• E.g. The Sum of Hearing Threshold Levels 
Shows an Increase of 30 dB Compared to 
Preceding Level. Very Serious Shift! 
(Case Study : Less than 10 yrs employment, 1st cluster 20 dB and 
below. 2nd cluster 30 dB and below. 
: More than 10yrs employment, cluster 30 dB and 
below) 
• International Standard Organisation (ISO)
Real Examples 
• Vibration Analysis 
Pumps or Compressors 
• Pump fails 
• Compressor showing high 
vibration (fitters showing 
concern of the compressor 
vibration) 
Frequency 
Amplitude 
f 2f 3f 
After correction
Counter Measures 
• Condition Monitoring Program 
- Compressors and Pumps Characteristics 
• Principle Of Vibration Is Caused By Rotation of Shaft 
and Components Connected to the Shaft Inside The 
Compressors Or Pumps, A Sinusoidal Behaviour 
Amplitude Y 
Time 
Direction 
X 
Direction 
Vibration in X 
and Y 
Direction 
Shaft 
Rotation 
Direction 
f = 1/T 
f = frequency 
T
Counter Measures 
• Identify Equipment to Measure, Critical Pumps 
or Compressors (e.g. Screw Compressor) 
• Collection of Data at Specified Points, e.g. X-Direction 
or Y-Direction 
• Storage of Data 
• Analyse and Compare Spectrum 
• Corrective and Rectification Actions, Take Your 
Time to Decide 
• Spot the Rise 
• DO NOT Exceed the Limit
Counter Measures 
• Engage a Vibration Consultant 
Collection Spot 
Frequency 
Amplitude 
f 2f 3f 
Analyse 
Data 
Collect 
Data 
Corrective 
Action 
Axial Compressor 
Invoice 
Pump Blade 
1 Qty 
……………. 
Received By 
Executive 
√
Counter Measures 
• Compressor was overhaul, monitored and 
shows low vibration. 
• Pump Performance before and after 
shows the peak has reduced. 
• To Avoid Failures Which Will Cause 
Process Interruption. If NO Breakdown, 
Smooth Operation
Real Examples 
• Reducing Machine Set Up Time – An 
Improvement in Production Output 
- Trending 
• The Set Up Time was Recorded, From Start Time to 
Finish Time
Real Examples 
SETUP TIME 
14 
12 
10 
8 
6 
4 
2 
0 
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 
NUMBER OF SETUPS 
TIME 
Target 
Actual Time 
Linear Trend Line 
Moving Average (of 
Past 5 Setup)
Real Examples 
• To Reduce Setup Time, Is To Identify Time Consuming 
Tasks 
• After Identifying The Time Consuming Tasks, Actions 
Have to be Taken To Reduce These Tasks, e.g. By 
Quick Fit, Work Continuously Along Critical Activities 
(take note of resource exhaustion), Adjust Machine Parts 
Before Set Up Activities (not directly connected to the 
machine), Re-design same activities, etc. 
• Monitor Time To Avoid Slippage, Inform Everybody 
• Plan Actions
Counter Measures 
• Time Motion Study 
- Determine Tasks to Monitor Progress 
- Concurrent Activities (Resource, Sequence) 
- Engineering Measures 
- Team Approach 
Maintenance 
Time 
Set Up 
Time 
Operation Time 
Total Time 
Reduced
Counter Measures 
• The Total Time 
- Identify Tasks 
• Number of 
Resources 
- Spot the Critical 
Activities 
(Including Bottle 
Neck) 
Task 1 
Task 2 
1 Man 
1 Man Task 3 
Finish 
Start 
1Man 
2 Men 
Start 1 Man Finish 
Total Time
Real Examples 
• Setup Time to be Reduced to Achieved 
Shorter Setup Time 
• More Production Flexibility In Some 
Application 
Product A 
Setup 
Product B 
Setup 
Product A, …
Known Knowledge 
• That There Will be Failures Due to Failure Rate. 
• Bath Tube Theory, Wear and Tear, Electronic 
Failure, etc. 
• Field Detectors Faults 
• Valve fail to Close 
• Drawing Representation/Specification Errors
Counter Measures 
• Redundancy 
- 1 Out Of 2 Voting (Logic) In The Field 
(In Some Application – 2 out of 3)
Counter Measures 
• Redundancy In The 
Field 
• A Third Detector In 
Some Projects To 
Detect The Shadow 
(triple!) 
• Same like pumps, 
one operating and 
one spare 
FD 1 
FD 2 
Shadow V1 
Area if 
FD1 fail 
FD 3 to detect 
hidden area
Counter Measures 
• Redundancy 
Throughout, i.e. 
From the Field Up to 
the Input/Output Card 
and Process 
Input 
Input 
Detector 
Detector 
Processor Output 
Processor Output
Counter Measures 
• Partial Stroke Test 
- Able to Detect 70% of Valve Failures 
• Full Stroke Test 
- Remaining 30% of Valve Failures 
Flow 
Direction 
open 
Partial 
close 
(p5, Partial-Stroke Testing of Block Valves, 
Angela E. Summers, Control Engineering, Nov 2000) 
Flow 
Direction 
open 
close
Counter Measures 
• Review / Check 
- Try to Spot The Error 
• By keep checking for error, there 
is a high possibility of reducing the 
errors. 
• Check technical contents, 
specified correctly, etc. 
• Quality guru will say that quality is 
driven to a certain extent only. 
After which it is worth it. 
• Also do not spent too much time 
checking, otherwise, waste of 
time, not efficient, non productive, 
etc. 
Start Review / 
Check 
IDC 
Review / 
Check 
IFR 
Review / 
Check 
AFC Finish (or 
As-Built) 
IDC Interdiscipl 
inary 
Check 
AFC Approved 
For 
Constructi 
on 
IFR Issued For 
Review
Conclusion 
• Identify The Failure(s) 
• Take Time to Study/Investigate (Collect Data, 
Analyse, Trending, etc.) 
• Ask Around for Available Practices/Advise 
• Take Action(s) after that, do not do it by yourself, 
make sure company management know about it. 
• Make Use of the Prediction Concepts and 
Proven Practices to Avoid Failures in Plant or Oil 
and Gas Facilities.
Conclusion 
• Areas that were looked into: 
- Plant Noise 
- Pumps and Compressors Vibration 
- Machine Setup Time 
- Redundancy 
- Partial/Full Stroke Testing 
- Review/Check Activity
Predicting Failures 
• Prediction is Possible to Resolved Failures 
• Prediction Requires Past Data 
• It is Scientific Based, i.e. Measurable 
Built Trend/Threshold Limit/Moving 
Average/Analysis, to Predict a Better 
Future. 
• Spot the Change (Before and After)
Predicting Failures 
Stages In Predicting Failure 
Identify Failure Study/Investigate Corrective 
Action 
Counter 
Measures 
Collect Data/ Records Threshold/Analyse/Tre 
nd/Moving Average 
PREDICTION 
Failure Rate
Speaker Contact 
• Current Employer : 
Technip Geoproduction (M) Sdn. Bhd. 
• Position : Instrument Engineer 
Any queries?

More Related Content

PPT
Ssoverview
PPT
Introduction to six sigma
PDF
Risk management in complex changeovers through cfmea
PPTX
TECO HK Series Microwave Consistency Transmitter Technical Review
PPT
SMED-Observation Training
PPT
How to reduce changeover loss
PPTX
SMED or Single Minute Exchange of Dies and Set-up Reduction Process
PPT
Island Of Excellence Presentation
Ssoverview
Introduction to six sigma
Risk management in complex changeovers through cfmea
TECO HK Series Microwave Consistency Transmitter Technical Review
SMED-Observation Training
How to reduce changeover loss
SMED or Single Minute Exchange of Dies and Set-up Reduction Process
Island Of Excellence Presentation

What's hot (10)

PDF
The SMED Method by CLT Services
PPTX
Cynergy general overview v2
PDF
Change your board operator to a process control manager
PPTX
Setup Reduction Workshop
PDF
High-Fidelity Operator Training Simulator for CCGT Implemented Before Plant C...
PPT
Safe work procedure1
PDF
How to Make Money with Your Operator Graphics
PDF
Project Proposal for Predictive Maintenance Power Plant
PDF
Preventive maintenance
PPT
Smed smpl 1
The SMED Method by CLT Services
Cynergy general overview v2
Change your board operator to a process control manager
Setup Reduction Workshop
High-Fidelity Operator Training Simulator for CCGT Implemented Before Plant C...
Safe work procedure1
How to Make Money with Your Operator Graphics
Project Proposal for Predictive Maintenance Power Plant
Preventive maintenance
Smed smpl 1
Ad

Viewers also liked (20)

PPTX
Condition monitoring of rotating machines ppt
PDF
Angle Type Control Valve Manufacturer in India
PDF
Understanding Machines Predicting machine failure
PDF
Comparison of analytical methods of analysing linear induction motors
PDF
Sensors Expo 2013: Condition Based Maintenance, Evigia Systems
PDF
IPSA-CERTUS-MAEPSA
PPTX
Total Productive Maintenance
PPTX
Implementation of condition based maintenance in manufacturing industry
PPTX
Condition based maintenance strategy
PPTX
Part 6 of 6 - Operation Phase - Safety Lifecycle Seminar - Emerson Exchange 2010
PDF
Fundamentals of vibration_measurement_and_analysis_explained
PDF
Bearing dianogsis 1
PDF
01 intro
PPTX
Chuyen de 3 a 1 bearings defects-n
PPT
revamps-reciprocating-compressor
PDF
PDF
Why-Why Analysis in Arabic Language.
PPTX
Good Vibrations: Measuring the Vibrations of Touch.
PDF
Frequency Response Analysis by Prof Satish IISc
PDF
Mechanical Vibrations
Condition monitoring of rotating machines ppt
Angle Type Control Valve Manufacturer in India
Understanding Machines Predicting machine failure
Comparison of analytical methods of analysing linear induction motors
Sensors Expo 2013: Condition Based Maintenance, Evigia Systems
IPSA-CERTUS-MAEPSA
Total Productive Maintenance
Implementation of condition based maintenance in manufacturing industry
Condition based maintenance strategy
Part 6 of 6 - Operation Phase - Safety Lifecycle Seminar - Emerson Exchange 2010
Fundamentals of vibration_measurement_and_analysis_explained
Bearing dianogsis 1
01 intro
Chuyen de 3 a 1 bearings defects-n
revamps-reciprocating-compressor
Why-Why Analysis in Arabic Language.
Good Vibrations: Measuring the Vibrations of Touch.
Frequency Response Analysis by Prof Satish IISc
Mechanical Vibrations
Ad

Similar to PREDICTING FAILURES_AUGUST2007 (20)

PPT
Process Control Improvement Primer - Greg McMillan Deminar
PPT
PDM IMPLIMENTATION
PPT
Isa saint-louis-exceptional-opportunities-short-course-day-1
DOC
jayakumarcv
PPT
PCE on rig for Engnineer and non eng information.ppt
DOC
Instrument engineer CV
PDF
Implémentation Méthode Single Minute Exchange of Dies(SMED)
PPTX
How to Calibrate a Pressure Gauge with a Pressure Comparator or Calibrator We...
PPTX
TPM2015 Presentation by Andre Gibson - Key Engineering Solutions
PPTX
0405process_instrumentation_and_control.pptx
PPT
16 major losses tng
PPTX
General guidelines for building commissioning ppt.pptx
PPTX
General guidelines for building commissioning ppt
PPTX
Learning Unit 5_PROCESS CONTROL.pptx
PPT
Process Capability.ppt related to quality
PDF
A practical approach to pharmaceutical HVAC energy reduction
PPTX
productivity improvement using work studt tools.pptx
PDF
Motion and time study
PPT
Fluke and Transcat Present: 4 Common Pressure Calibration Pain Points Webinar
PDF
Designing the Enterprise for Manufacturing
Process Control Improvement Primer - Greg McMillan Deminar
PDM IMPLIMENTATION
Isa saint-louis-exceptional-opportunities-short-course-day-1
jayakumarcv
PCE on rig for Engnineer and non eng information.ppt
Instrument engineer CV
Implémentation Méthode Single Minute Exchange of Dies(SMED)
How to Calibrate a Pressure Gauge with a Pressure Comparator or Calibrator We...
TPM2015 Presentation by Andre Gibson - Key Engineering Solutions
0405process_instrumentation_and_control.pptx
16 major losses tng
General guidelines for building commissioning ppt.pptx
General guidelines for building commissioning ppt
Learning Unit 5_PROCESS CONTROL.pptx
Process Capability.ppt related to quality
A practical approach to pharmaceutical HVAC energy reduction
productivity improvement using work studt tools.pptx
Motion and time study
Fluke and Transcat Present: 4 Common Pressure Calibration Pain Points Webinar
Designing the Enterprise for Manufacturing

More from Gan Chun Chet (20)

PPTX
The Mutant Mint Leaf - A DNA Replication (Aquaponics System)
PDF
Machine Setup Time - A Limit Determiner (Tool)
PPTX
Trust Measurement Presentation_Part 3
PDF
Predicting Machine Setup Time_Suggeting A Limit Determiner, Institute of Engi...
PDF
Predictive Method in Analyzing Past Data_(May_09), Institute of Engineers Mal...
PDF
Modelling Trust of Customer and Supplier Interaction_2023 Vol 10 Issue 1.pdf
PDF
Oil and Gas Industry : Datasheet for Fire and Gas Detectors.pdf
PPT
Fire and Gas Detection System : Part 3_Technical Features, Locating Detectors...
PPT
Fire and Gas Detection System : Part 2_Block Diagram_Philosophy, Signal Types...
PPT
Fire and Gas Detection System : Part 1_The Field Devices and Its Panels
PPTX
Trust Measurement Presentation_2.pptx
PPTX
Boundary Layer Oscillation for Aircraft Cavity/Surfaces
PPTX
Noise dB to dBA Calculation_Gan Chun Chet.pptx
PPTX
Noise Error Calculation_Gan Chun Chet.pptx
PDF
Noise_Proofing Linking Equations.pdf
PDF
The Prediction of Time Trending Techniques in Manufacuturing
DOCX
The Philosophy and Characteristics of Trust Measurement in Practice.docx
PDF
Trust Measurement Gan Chun Chet 07.01.2022
PDF
mV to dBm Airflow Across Cavity_Gan Chun Chet_07.01.2021
PDF
Control System_3rd Order System_Root Locus Method
The Mutant Mint Leaf - A DNA Replication (Aquaponics System)
Machine Setup Time - A Limit Determiner (Tool)
Trust Measurement Presentation_Part 3
Predicting Machine Setup Time_Suggeting A Limit Determiner, Institute of Engi...
Predictive Method in Analyzing Past Data_(May_09), Institute of Engineers Mal...
Modelling Trust of Customer and Supplier Interaction_2023 Vol 10 Issue 1.pdf
Oil and Gas Industry : Datasheet for Fire and Gas Detectors.pdf
Fire and Gas Detection System : Part 3_Technical Features, Locating Detectors...
Fire and Gas Detection System : Part 2_Block Diagram_Philosophy, Signal Types...
Fire and Gas Detection System : Part 1_The Field Devices and Its Panels
Trust Measurement Presentation_2.pptx
Boundary Layer Oscillation for Aircraft Cavity/Surfaces
Noise dB to dBA Calculation_Gan Chun Chet.pptx
Noise Error Calculation_Gan Chun Chet.pptx
Noise_Proofing Linking Equations.pdf
The Prediction of Time Trending Techniques in Manufacuturing
The Philosophy and Characteristics of Trust Measurement in Practice.docx
Trust Measurement Gan Chun Chet 07.01.2022
mV to dBm Airflow Across Cavity_Gan Chun Chet_07.01.2021
Control System_3rd Order System_Root Locus Method

PREDICTING FAILURES_AUGUST2007

  • 1. PREDICTING FAILURES Ir. Gan Chun Chet August 2007
  • 2. The Concept and Practice • In the Meeting Room Figures Are Showing Better Outlook • The Ability to Know Bad Results Before It Happens, Would Be Good to Know Before It Occurs • Hence, A Scientific Method to Measure Past Records to Achieve the Required Future Targets Emerge I have done enough. Engineer Everything Seems to Be OK. Plant Manager Profit is Rising I’m Impress Managing Director PREDICTING FAILURES
  • 3. Predicting Failures • What is Predicting? • The ability to know something before it happens • What are Failures? • Something that does not pass the required results
  • 4. Predicting Failures • Some items that are highlighted in this talk - Plant Noise - Pumps and Compressors Vibration - Machine Setup Time - Redundancy - Partial/Full Stroke Testing - Review/Check Activity
  • 5. Predicting Failures • Introduction • Passes, Failures and Predictions • Some Examples To Serve as A Guide • Some Counter Measures to Overcome • Conclusion
  • 6. Introduction • Practices in General Industry Engineering Project and Maintenance Services - Engineering Best Practices • Oil and Gas Consultancy and Contracting Services - Requires High Safety Standards - Requires High Reliable Products
  • 7. Introduction • Some of the Issues Addressed in General Industry (Manufacturing and Processing) - Hearing Losses - Pumps and Compressors Concerns of A Breakdown - Long Machine Setup Time Which Contributed to Uneasiness Among Production Personnel (Production Planning, Production Morale, Machine Change Over, etc.) • Some of the Practices from the Oil and Gas Industry that worth while knowing - Redundancy - Partial/Full Stroke Testing - Review Check Activities
  • 8. General Industry - Engineering Project and Maintenance Services • Improve Plant Conditions - Safe Working Areas Inhabitants in the plant are able to work in a healthy and conducive environment • Reduce Breakdowns - Continuous Batch Operation The production purpose is to ensure continuous supply to the market due to demand (to operate profitably)
  • 9. Oil and Gas - Consultancy and Contracting Services • Query Services - Latest Development Client would like to know whether the latest product is purchased • Design and Engineering Services - Specify the Right Requirements - Integrating Properly Engineered correctly to enable the platform or plant operating within the design production output, a proper detection system, the best way to fight a fire, equipment are maintain to it best performance, etc.
  • 10. What Causes Failures? • Causes of Failures - Wear and Tear (Bath Tub Theory) - Vibration - Exhaustive Operation • Are Failures Detectable? - Can it be Spotted and Rectified? - Are there Methods to Detect Failures? (i.e. To know before something happen)
  • 11. Passes, Failures and Predictions • 100% Passes - Is This Achievable (Take note of Process Variance) • 0% Failures - Is this achievable (Perfection? – Good Engineering Design) • T+1 Prediction - Based on Facts and Past Data
  • 12. Passes • National or University Examination Results - Number of Students with Distinction Is increasing the passing rate an indication of better quality students? University in the search of glooming student that are able to meet today’s challenges • Productivity - Operation Output The measurement of productivity is to achieve better result each year until it reaches its best performance. Business/Management will begin to look for other performance parameters
  • 13. Passes • Upper Range Limit - Highest Acceptable Specification Value • Lower Range Limit - Lowest Acceptable Specification Value 120 100 80 60 • Is it possible to increase the 40 limit, hence more will pass • Take note of design quality 20 • Take note of process (machine) limit 0 LRL URL 0 5 10 15
  • 14. Failures • Bath Tube Theory - Failures Occur Beginning of Design Life. - Sporadic Failures During Normal Life Stage. - Failures Begins to Occur Towards the End of Design Life. • Is This Actually Happening in Real Operation? 7 6 5 4 3 2 1 0 0 10 20 Time Failure Rate Infant Mortality Stage Wear-out Stage Normal-life Stage
  • 15. Failures • Off-Spec Rejects - Not Meeting Specification - A Known Process Variation • Sporadic Peaks - Sudden Occurrences - Unpredictable 150 100 50 0 300 200 100 0 1 2 3 4 5 6 7 8 9 1
  • 16. Failures • Shutdown/Detection System Fail To Operate - Not Able to Detect or Shutdown • Valves Unable to Function - Stuck - Cannot Close
  • 17. Failures • Failure Rate = Number of Failures X 100 Total Number of Products Tested (p776, Operations Management – Nigel Slack et al, Pitman Publishing) • An equation to know the number of actual failure coming out of the production line • Measurement of the percentage of product failure
  • 18. Failures • Mean Time Between Failure (MTBF) = Operating Hours Number of Failures (p776) The average time between failures, measured in actual operation. The number of breakdowns over the operation duration. (To Determine Time Before a Failure, Average of Past Record)
  • 19. Prediction • Threshold Limit - Above a Set Horizontal Line Will Trigger a Warning. • Review of past data shows that a limit has to be set. Anything above this set limit must need to be attended. • E.g. Noise (Hearing) Level 8 7 6 5 4 3 2 1 0 Threshold Limit Warning 0 5 10 15
  • 20. Prediction • Analysis - Frequency Spectrum • An analysis of the pump or compressor vibration behaviour. Because the pump and compressor rotate at fixed angular rotation, the frequency of the pump or compressor can be calculated. The spectrum shows the frequency of the pump or compressor, etc. Frequency Amplitude f 2f 3f
  • 21. Prediction • Trend Line - A Straight Line Can Be Drawn on The Collected Data. - The Straight Line Can be Extended to Know the Next Value. E.g. A Linear Relationship Between X and Y and be Concluded, Case of reduction in setup time. Line of Regression 10 5 0 0 5 10
  • 22. Prediction • Moving Average - Average of Past Data and Moving Forward • The moving average line of 2 past data is the average of 2 past data. • The moving average line of 2 past data is more reactive than 5 past data. • Moving average is the record of past 90 80 70 60 50 40 30 historical data. With this type of trend line, the future up or down trend can 20 be know, provided all factors remain unchange. 10 (E.g. Monitoring Setup time) 0 5 past data 2 past data 0 5 10 15
  • 23. Real Examples • Threshold Shift (Noise Study) – (Factories and Machinery, Noise Exposure) Regulations, 1989 (A Real Noise Reduction Project) • Noise Level Collected and Recorded at 500, 1000, 2000, 3000 & 4000Hz • Sampled of Production Workers at Floor Level
  • 24. Real Examples • Less than 10 years service • 2 Clusters • Results - Below 20 dB (Cluster 1) - Below 30 dB (Cluster 2)
  • 25. Real Examples • More than 10 years service • 1 Cluster • Results - Below 30 dB
  • 26. Counter Measures • Hearing Conservation Program - Engineering Control (To Install Enclosure At Rotating Equipment, To Reduce Device Noise e.g. Control Valve & Relief Valve, To Prevent Noise From Traveling Into The Control Room) Enclosure Machine Limit noise out of the enclosure Control Room Limit noise to the control room
  • 27. Counter Measures • - Administrative Control (To Move Workers In And Out Of Noisy Area (take note of threshold shift, very dangerous, so act mentioned to limit daily exposure to noise by control of work schedule, very difficult to practice), Constant Monitoring, Do Not Allow People To Enter Noisy Areas, Sign Board of High Noise Areas, etc.) HIGH NOISE AREA - WEAR EAR MUFFLE
  • 28. Counter Measures • Required by law that the Permissible Exposure Limit is - 85 dB(A) for 16 hours duration per day, - 90 dB(A) for 8 hours duration per day, etc per schedule - Nobody shall be exposured to noise level exceeding 115 dB (A) at any time Factories And Machinery (Noise Exposure) Regulations, 1989 • Employees need to check for a baseline reading (required by law) • Take note of Threshold Shift! DO NOT play with this! • Reduce Noise Level (Engineering Measures) • Consult Noise Consultant or Ear Specialist!
  • 29. Counter Measures • Some Examples (Also Consult Doctors or Specialist! Hearing Problem Must not be Taken Lightly) • E.g. The Sum of Hearing Threshold Levels Shows an Increase of 30 dB Compared to Preceding Level. Very Serious Shift! (Case Study : Less than 10 yrs employment, 1st cluster 20 dB and below. 2nd cluster 30 dB and below. : More than 10yrs employment, cluster 30 dB and below) • International Standard Organisation (ISO)
  • 30. Real Examples • Vibration Analysis Pumps or Compressors • Pump fails • Compressor showing high vibration (fitters showing concern of the compressor vibration) Frequency Amplitude f 2f 3f After correction
  • 31. Counter Measures • Condition Monitoring Program - Compressors and Pumps Characteristics • Principle Of Vibration Is Caused By Rotation of Shaft and Components Connected to the Shaft Inside The Compressors Or Pumps, A Sinusoidal Behaviour Amplitude Y Time Direction X Direction Vibration in X and Y Direction Shaft Rotation Direction f = 1/T f = frequency T
  • 32. Counter Measures • Identify Equipment to Measure, Critical Pumps or Compressors (e.g. Screw Compressor) • Collection of Data at Specified Points, e.g. X-Direction or Y-Direction • Storage of Data • Analyse and Compare Spectrum • Corrective and Rectification Actions, Take Your Time to Decide • Spot the Rise • DO NOT Exceed the Limit
  • 33. Counter Measures • Engage a Vibration Consultant Collection Spot Frequency Amplitude f 2f 3f Analyse Data Collect Data Corrective Action Axial Compressor Invoice Pump Blade 1 Qty ……………. Received By Executive √
  • 34. Counter Measures • Compressor was overhaul, monitored and shows low vibration. • Pump Performance before and after shows the peak has reduced. • To Avoid Failures Which Will Cause Process Interruption. If NO Breakdown, Smooth Operation
  • 35. Real Examples • Reducing Machine Set Up Time – An Improvement in Production Output - Trending • The Set Up Time was Recorded, From Start Time to Finish Time
  • 36. Real Examples SETUP TIME 14 12 10 8 6 4 2 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 NUMBER OF SETUPS TIME Target Actual Time Linear Trend Line Moving Average (of Past 5 Setup)
  • 37. Real Examples • To Reduce Setup Time, Is To Identify Time Consuming Tasks • After Identifying The Time Consuming Tasks, Actions Have to be Taken To Reduce These Tasks, e.g. By Quick Fit, Work Continuously Along Critical Activities (take note of resource exhaustion), Adjust Machine Parts Before Set Up Activities (not directly connected to the machine), Re-design same activities, etc. • Monitor Time To Avoid Slippage, Inform Everybody • Plan Actions
  • 38. Counter Measures • Time Motion Study - Determine Tasks to Monitor Progress - Concurrent Activities (Resource, Sequence) - Engineering Measures - Team Approach Maintenance Time Set Up Time Operation Time Total Time Reduced
  • 39. Counter Measures • The Total Time - Identify Tasks • Number of Resources - Spot the Critical Activities (Including Bottle Neck) Task 1 Task 2 1 Man 1 Man Task 3 Finish Start 1Man 2 Men Start 1 Man Finish Total Time
  • 40. Real Examples • Setup Time to be Reduced to Achieved Shorter Setup Time • More Production Flexibility In Some Application Product A Setup Product B Setup Product A, …
  • 41. Known Knowledge • That There Will be Failures Due to Failure Rate. • Bath Tube Theory, Wear and Tear, Electronic Failure, etc. • Field Detectors Faults • Valve fail to Close • Drawing Representation/Specification Errors
  • 42. Counter Measures • Redundancy - 1 Out Of 2 Voting (Logic) In The Field (In Some Application – 2 out of 3)
  • 43. Counter Measures • Redundancy In The Field • A Third Detector In Some Projects To Detect The Shadow (triple!) • Same like pumps, one operating and one spare FD 1 FD 2 Shadow V1 Area if FD1 fail FD 3 to detect hidden area
  • 44. Counter Measures • Redundancy Throughout, i.e. From the Field Up to the Input/Output Card and Process Input Input Detector Detector Processor Output Processor Output
  • 45. Counter Measures • Partial Stroke Test - Able to Detect 70% of Valve Failures • Full Stroke Test - Remaining 30% of Valve Failures Flow Direction open Partial close (p5, Partial-Stroke Testing of Block Valves, Angela E. Summers, Control Engineering, Nov 2000) Flow Direction open close
  • 46. Counter Measures • Review / Check - Try to Spot The Error • By keep checking for error, there is a high possibility of reducing the errors. • Check technical contents, specified correctly, etc. • Quality guru will say that quality is driven to a certain extent only. After which it is worth it. • Also do not spent too much time checking, otherwise, waste of time, not efficient, non productive, etc. Start Review / Check IDC Review / Check IFR Review / Check AFC Finish (or As-Built) IDC Interdiscipl inary Check AFC Approved For Constructi on IFR Issued For Review
  • 47. Conclusion • Identify The Failure(s) • Take Time to Study/Investigate (Collect Data, Analyse, Trending, etc.) • Ask Around for Available Practices/Advise • Take Action(s) after that, do not do it by yourself, make sure company management know about it. • Make Use of the Prediction Concepts and Proven Practices to Avoid Failures in Plant or Oil and Gas Facilities.
  • 48. Conclusion • Areas that were looked into: - Plant Noise - Pumps and Compressors Vibration - Machine Setup Time - Redundancy - Partial/Full Stroke Testing - Review/Check Activity
  • 49. Predicting Failures • Prediction is Possible to Resolved Failures • Prediction Requires Past Data • It is Scientific Based, i.e. Measurable Built Trend/Threshold Limit/Moving Average/Analysis, to Predict a Better Future. • Spot the Change (Before and After)
  • 50. Predicting Failures Stages In Predicting Failure Identify Failure Study/Investigate Corrective Action Counter Measures Collect Data/ Records Threshold/Analyse/Tre nd/Moving Average PREDICTION Failure Rate
  • 51. Speaker Contact • Current Employer : Technip Geoproduction (M) Sdn. Bhd. • Position : Instrument Engineer Any queries?