jan_eite.bullema@tno.nl
Bayesian Networks
to analyse LED reliability
An introduction
Jan Eite Bullema – ASQ Certified Reliability Engineer
jan_eite.bullema@tno.nl
Outline
Bayesian Networks for Analysis of LED reliability
Bayesian Networks for analysis of LED reliability
2
Overview of Reliability Science
What are Bayesian Networks?
Challenges in LED System complexity
Limitations in part based lifetime predictions
Modern reliability approaches,
e.g. MOEST, Bayesian Networks, Big Data
Conclusions
jan_eite.bullema@tno.nl
What is Reliability
Definitions of Quality and Reliability
Bayesian Networks for analysis of LED reliability
3
1996, Lewis
Quality
The ability of a product to fulfil its intended purpose
Reliability
The ability of a product to fulfil its intended purpose
for a certain period of time under stated conditions
jan_eite.bullema@tno.nl
Why is Reliability Important?
Business Issues
Bayesian Networks for analysis of LED reliability
4
---------------
Reputation
Customer Satisfaction
Warranty Costs
Repeat Business
Cost Analysis
Customer Requirements
Competitive Advantage
jan_eite.bullema@tno.nl
The bath-tub curve
Reliability Life-Model
Bayesian Networks for analysis of LED reliability
5
Source: R. Crowe, Design for Reliability, CRC Press 2001
jan_eite.bullema@tno.nl
What is a Bayesian Network?
Bayesian Networks are based upon Bayes Theorem
Bayesian Networks for analysis of LED reliability
6
Source: Bayesian Statistics – a brief introduction, Ken Rice, 2014
Common sense reduced to computation
Pierre-Simon, marquis de Laplace (1749-1827)
Inventor of Bayesian inference
jan_eite.bullema@tno.nl
What is a Bayesian Network?
A Bayesian Network is a Directed Acyclic Graph
Bayesian Networks for analysis of LED reliability
7
Source: TNO – report Dynamic Bayesian Network simulations of a lighting system
jan_eite.bullema@tno.nl
What is a Bayesian Network?
Bayesian Network built from Expert Knowledge
Step by Step Example for a simple LED system
Bayesian Networks for analysis of LED reliability
8
Source: TNO – report Dynamic Bayesian Network simulations of a lighting system
Step 1. System Decomposition
jan_eite.bullema@tno.nl
What is a Bayesian Network?
Bayesian Network built from Expert Knowledge
Step 2. Failure Mode and Effect Analysis
Bayesian Networks for analysis of LED reliability
9
Source: TNO – report Dynamic Bayesian Network simulations of a lighting system
jan_eite.bullema@tno.nl
What is a Bayesian Network?
Bayesian Network built from Expert Knowledge
Step 3. Failure Tree from FMEA
Bayesian Networks for analysis of LED reliability
10
Source: TNO – report Dynamic Bayesian Network simulations of a lighting system
jan_eite.bullema@tno.nl
What is a Bayesian Network?
Bayesian Network built from Expert Knowledge
Step 4. Translate Failure Tree for computation
Bayesian Networks for analysis of LED reliability
11
Source: TNO
Step 5. Calculate Physics of Failure
A_SOL_1 <- exp(Ea_SOL/KB*(1/TE-1/T_SOL_1))
R_SOL_1 <- exp(-TIME * A_SOL_1 * (10^9/FIT_SOL)^-1)
Step 6. Do a Monte Carlo Simulation
jan_eite.bullema@tno.nl
What is a Bayesian Network?
Bayesian Network built from Expert Knowledge
Step 7. Do Bayesian Inference
Bayesian Networks for analysis of LED reliability
12
Source: TNO
Step 8. Calculate Reliability Predictions
jan_eite.bullema@tno.nl
What is a Bayesian Network?
Bayesian Network built from Expert Knowledge
Step 9. Repeat steps 5 - 8 over a time window
Bayesian Networks for analysis of LED reliability
13
Source: TNO
Step 10. Analyse Reliability Predictions
jan_eite.bullema@tno.nl
Outline
Bayesian Networks for Analysis of LED reliability
Bayesian Networks for analysis of LED reliability
14
Overview of Reliability Science
What are Bayesian Networks?
Challenges in LED reliability
i.e. Lifetimes > 50.000 hrs.
Modern reliability approaches,
e.g. MOEST, Bayesian Networks, Big Data
Conclusions
jan_eite.bullema@tno.nl
Reliability Basics
What is the reliability of a single LED
Bayesian Networks for analysis of LED reliability
15
http://rl.omslighting.com/ledacademy/694/led-academy
LED manufacturer claims that
50% of their LEDs (from any
batch) will emit at least 70% of
the initial lumens after 50,000
hours of operation
jan_eite.bullema@tno.nl
Reliability Basics
What is the reliability of a single LED
E.g. the Mean Time Between Failures
Bayesian Networks for analysis of LED reliability
16
ETAP White Paper: LED Lifetime in Practice
jan_eite.bullema@tno.nl
Reliability Basics
What is the reliability of a single LED
e.g. Temperature Dependence of MTBF
Bayesian Networks for analysis of LED reliability
17
LumiLED High Flux High Power Reliability Data
jan_eite.bullema@tno.nl
Reliability Basics
What is the reliability of a single LED?
e.g. Temperature Dependence of MTBF
Bayesian Networks for analysis of LED reliability
18
LumiLED High Flux High Power Reliability Data
jan_eite.bullema@tno.nl
Durability of LED Systems
Lifetimes > 50.000 hrs and Mission Profiles
Bayesian Networks for analysis of LED reliability
19
Life time of a LED System has to be 25 000 – 50 000 hrs.
Gielen et al, Development of an intelligent integrated LED system-in-package, EPMC 2011
jan_eite.bullema@tno.nl
Durability of LED Systems
System behaviour is more than the sum of
component behaviour
Bayesian Networks for analysis of LED reliability
20
http://www.ledsmagazine.com
jan_eite.bullema@tno.nl
LED Systems are Complex
Knowing the behaviour of parts is not sufficient
Bayesian Networks for analysis of LED reliability
21
Bullema, Combination of Bayesian Networks
and FEM models to Predict Reliability of LED Systems, ECTS 2012
Example: The ESIP Demonstrator:
jan_eite.bullema@tno.nl
LED Systems are Complex
Knowing the behaviour of parts is not sufficient
Bayesian Networks for analysis of LED reliability
22
Bullema, Combination of Bayesian Networks
and FEM models to Predict Reliability of LED Systems, ECTS 2012
Example: The CSSL Demonstrator:
jan_eite.bullema@tno.nl
LED Systems are Complex
Knowing the behaviour of parts is not sufficient
Bayesian Networks for analysis of LED reliability
23
Bullema, Combination of Bayesian Networks
and FEM models to Predict Reliability of LED Systems, ECTS 2012
Example: The ESIP Demonstrator:
compact LED system with integrated driver and string of six LEDs
LED
die
Driver
chip
Thermal
Pad
IO leads
Silicone
lens/filler
Moulding
compound
LED
die
Driver
chip
Thermal
Pad
IO leads
Silicone
lens/filler
Moulding
compound
jan_eite.bullema@tno.nl
LED Systems are Complex
Knowing the behaviour of parts is not sufficient
Bayesian Networks for analysis of LED reliability
24
Bullema, Combination of Bayesian Networks
and FEM models to Predict Reliability of LED Systems, ECTS 2012
Example: The ESIP Demonstrator:
jan_eite.bullema@tno.nl
LED Systems are Complex
Knowing the behaviour of parts is not sufficient
Bayesian Networks for analysis of LED reliability
25
Bullema, Combination of Bayesian Networks
and FEM models to Predict Reliability of LED Systems, ECTS 2012
Example: The ESIP Demonstrator:
Failure tree for electrical defects and reduced lumen output
jan_eite.bullema@tno.nl
LED Systems are Complex
Knowing the behaviour of parts is not sufficient
Bayesian Networks for analysis of LED reliability
26
Bullema, Combination of Bayesian Networks
and FEM models to Predict Reliability of LED Systems, ECTS 2012
Example: The ESIP Demonstrator:
Calculation R(t) for not cooled and heat pipe cooled (T = T not cooled - 20 C) Design
ESIP with and
without heat pipe
jan_eite.bullema@tno.nl
LED Systems are Complex
Knowing the behaviour of parts is not sufficient
Bayesian Networks for analysis of LED reliability
27
Bullema, Combination of Bayesian Networks
and FEM models to Predict Reliability of LED Systems, ECTS 2012
Example: The ESIP Demonstrator:
Predicted Survival Rate R(t) of LV @ 25 ºC
Continuous,1xSwitch/day,10xSwitch/day
Green line = 10 x switching / day
Red line = 1 x switched / day
Blue line = No switching
jan_eite.bullema@tno.nl
LED Systems are Complex
Knowing the behaviour of parts is not sufficient
Bayesian Networks for analysis of LED reliability
28
Bullema, Combination of Bayesian Networks
and FEM models to Predict Reliability of LED Systems, ECTS 2012
Example: The ESIP Demonstrator:
ESIP with and
without heat pipe
CSSL Design A, B
At Low and High Temp
jan_eite.bullema@tno.nl
LED Systems are Complex
Knowing the behaviour of parts is not sufficient
Use Mission Profiles to Predict Reliability
Bayesian Networks for analysis of LED reliability
29
Source: TNO report
jan_eite.bullema@tno.nl
Outline
Bayesian Networks for Analysis of LED reliability
Bayesian Networks for analysis of LED reliability
30
-------------------------
Overview of Reliability Science
What are Bayesian Networks?
Challenges in LED reliability
i.e. Lifetimes > 50.000 hrs.
Modern reliability approaches,
e.g. MOEST, Bayesian Networks, Big Data
Conclusions
jan_eite.bullema@tno.nl
MEOST: a jump into the future
Multiple Over Stress Testing: finding the weak spot
Bayesian Networks for analysis of LED reliability
31
Keki R. Bhote , World Class Reliability, ISBN 0-8144-0792-7
Various stresses and their action
jan_eite.bullema@tno.nl
The Future lies in use of Big Data
Measuring all lamps real time during using
advanced prognostics: all the tools are available
Bayesian Networks for analysis of LED reliability
32
http://www.smartindustry.nl/
jan_eite.bullema@tno.nl
Approach to develop Reliability Model based
upon a Bayesian Network
Bayesian Networks for analysis of LED reliability
33
Bullema, Combination of Bayesian Networks and FEM models to Predict Reliability of LED Systems, 2012

More Related Content

PPTX
2015 Reliability of complex systems
PPT
Myths and Legends of Microinverters and Other DC Power Converters for Solar E...
PDF
The Board Designer s Guide to Testable Logic Circuits Electronic Systems Engi...
PPT
IES_SD_led stds_b-2
PPTX
Producing current from solar panell.pptx
PDF
(Ebook) The Board Designer's Guide to Testable Logic Circuits by Colin Maunde...
PDF
Wireless Project Using CAPsMAN
PDF
(B) pepco energy solutions overview document email
2015 Reliability of complex systems
Myths and Legends of Microinverters and Other DC Power Converters for Solar E...
The Board Designer s Guide to Testable Logic Circuits Electronic Systems Engi...
IES_SD_led stds_b-2
Producing current from solar panell.pptx
(Ebook) The Board Designer's Guide to Testable Logic Circuits by Colin Maunde...
Wireless Project Using CAPsMAN
(B) pepco energy solutions overview document email

Similar to 2016 Bayesian networks to analyse led reliability (20)

PDF
The Board Designer s Guide to Testable Logic Circuits Electronic Systems Engi...
PPT
Right Management Solar Presentation
PPSX
Contribution to the optimization of energy withdrawn from a PV panel using an...
PDF
Lect2 up010 (100324)
PPT
GloSho'14: Company Showcase - Eutricity
PPTX
Study on reliability optimization problem of computer By Dharmendra Singh[Srm...
PDF
Introduction To Semiconductor Devices For Computing And Telecommunications Ap...
PPT
Victor ppt
 
PDF
Survey of energy-efficient solutions in Network
PDF
APsystems presentation: a global leader of micro inverter, RSD and MLPE products
PDF
PV –Fed Led Lighting Systems
PDF
IEC 61850 Technical Overview.pdf
PPT
09.50 Ernst Vrolijks
PDF
Design of Low Cost Line Impedance Stabilization Network Using RLC Components ...
PDF
Final year project ideas for electrical engineering eepowerschool.com
PDF
LumenCACHE Proposal Print 102916 1332
PDF
NCTu DIC 2012 term report
PDF
Analysis Guide for Electronics / Electrical Product Designers
PDF
IRJET-Hardware Implementation of VLC using CMOS Camera
The Board Designer s Guide to Testable Logic Circuits Electronic Systems Engi...
Right Management Solar Presentation
Contribution to the optimization of energy withdrawn from a PV panel using an...
Lect2 up010 (100324)
GloSho'14: Company Showcase - Eutricity
Study on reliability optimization problem of computer By Dharmendra Singh[Srm...
Introduction To Semiconductor Devices For Computing And Telecommunications Ap...
Victor ppt
 
Survey of energy-efficient solutions in Network
APsystems presentation: a global leader of micro inverter, RSD and MLPE products
PV –Fed Led Lighting Systems
IEC 61850 Technical Overview.pdf
09.50 Ernst Vrolijks
Design of Low Cost Line Impedance Stabilization Network Using RLC Components ...
Final year project ideas for electrical engineering eepowerschool.com
LumenCACHE Proposal Print 102916 1332
NCTu DIC 2012 term report
Analysis Guide for Electronics / Electrical Product Designers
IRJET-Hardware Implementation of VLC using CMOS Camera
Ad

More from Jan Eite Bullema (17)

PDF
2018 Example of a Digital Twin for 3 D printing
PPTX
2017 Electrical interconnects in miro fluidics
PPTX
2012 Biocompatibele MEMS / Microsystems Packaging
PPTX
2018 Reliability in the age of big data
PPTX
2017 3D Printing: stop prototyping, start producing!
PPTX
2011 Introduction micro and nanotechnology
PPTX
2017 Accelerated Testing: ALT, HALT and MEOST
PPTX
2016 Deep Learning with R and h2o
PPT
2007 Introduction MEOST
PPT
2015 Deep learning and fuzzy logic
PPTX
2015 3D Printing for microfluidics manufacturing
PPTX
2016 3D printing for organ on a chip
PPTX
2012 Introduction wire bonding
PPTX
2014 2D and 3D printing to realize innovative electronic products
PPTX
2014 Medical applications of Micro and Nano Technologies
PPTX
2016 How to make big data productive in semicon manufacturing
PPTX
2012 Reliable and Durable Micro Joining
2018 Example of a Digital Twin for 3 D printing
2017 Electrical interconnects in miro fluidics
2012 Biocompatibele MEMS / Microsystems Packaging
2018 Reliability in the age of big data
2017 3D Printing: stop prototyping, start producing!
2011 Introduction micro and nanotechnology
2017 Accelerated Testing: ALT, HALT and MEOST
2016 Deep Learning with R and h2o
2007 Introduction MEOST
2015 Deep learning and fuzzy logic
2015 3D Printing for microfluidics manufacturing
2016 3D printing for organ on a chip
2012 Introduction wire bonding
2014 2D and 3D printing to realize innovative electronic products
2014 Medical applications of Micro and Nano Technologies
2016 How to make big data productive in semicon manufacturing
2012 Reliable and Durable Micro Joining
Ad

Recently uploaded (20)

PDF
Applications of Equal_Area_Criterion.pdf
PPTX
Feature types and data preprocessing steps
PDF
Java Basics-Introduction and program control
PDF
First part_B-Image Processing - 1 of 2).pdf
PPTX
Petroleum Refining & Petrochemicals.pptx
PPTX
Amdahl’s law is explained in the above power point presentations
PDF
August -2025_Top10 Read_Articles_ijait.pdf
PPTX
ASME PCC-02 TRAINING -DESKTOP-NLE5HNP.pptx
PPTX
Measurement Uncertainty and Measurement System analysis
PPTX
Management Information system : MIS-e-Business Systems.pptx
PPTX
AUTOMOTIVE ENGINE MANAGEMENT (MECHATRONICS).pptx
PPTX
CN_Unite_1 AI&DS ENGGERING SPPU PUNE UNIVERSITY
PPTX
tack Data Structure with Array and Linked List Implementation, Push and Pop O...
PDF
Abrasive, erosive and cavitation wear.pdf
PDF
UEFA_Embodied_Carbon_Emissions_Football_Infrastructure.pdf
PPT
Chapter 1 - Introduction to Manufacturing Technology_2.ppt
PPTX
"Array and Linked List in Data Structures with Types, Operations, Implementat...
PDF
Unit1 - AIML Chapter 1 concept and ethics
PPTX
CyberSecurity Mobile and Wireless Devices
PPTX
wireless networks, mobile computing.pptx
Applications of Equal_Area_Criterion.pdf
Feature types and data preprocessing steps
Java Basics-Introduction and program control
First part_B-Image Processing - 1 of 2).pdf
Petroleum Refining & Petrochemicals.pptx
Amdahl’s law is explained in the above power point presentations
August -2025_Top10 Read_Articles_ijait.pdf
ASME PCC-02 TRAINING -DESKTOP-NLE5HNP.pptx
Measurement Uncertainty and Measurement System analysis
Management Information system : MIS-e-Business Systems.pptx
AUTOMOTIVE ENGINE MANAGEMENT (MECHATRONICS).pptx
CN_Unite_1 AI&DS ENGGERING SPPU PUNE UNIVERSITY
tack Data Structure with Array and Linked List Implementation, Push and Pop O...
Abrasive, erosive and cavitation wear.pdf
UEFA_Embodied_Carbon_Emissions_Football_Infrastructure.pdf
Chapter 1 - Introduction to Manufacturing Technology_2.ppt
"Array and Linked List in Data Structures with Types, Operations, Implementat...
Unit1 - AIML Chapter 1 concept and ethics
CyberSecurity Mobile and Wireless Devices
wireless networks, mobile computing.pptx

2016 Bayesian networks to analyse led reliability

  • 1. jan_eite.bullema@tno.nl Bayesian Networks to analyse LED reliability An introduction Jan Eite Bullema – ASQ Certified Reliability Engineer
  • 2. jan_eite.bullema@tno.nl Outline Bayesian Networks for Analysis of LED reliability Bayesian Networks for analysis of LED reliability 2 Overview of Reliability Science What are Bayesian Networks? Challenges in LED System complexity Limitations in part based lifetime predictions Modern reliability approaches, e.g. MOEST, Bayesian Networks, Big Data Conclusions
  • 3. jan_eite.bullema@tno.nl What is Reliability Definitions of Quality and Reliability Bayesian Networks for analysis of LED reliability 3 1996, Lewis Quality The ability of a product to fulfil its intended purpose Reliability The ability of a product to fulfil its intended purpose for a certain period of time under stated conditions
  • 4. jan_eite.bullema@tno.nl Why is Reliability Important? Business Issues Bayesian Networks for analysis of LED reliability 4 --------------- Reputation Customer Satisfaction Warranty Costs Repeat Business Cost Analysis Customer Requirements Competitive Advantage
  • 5. jan_eite.bullema@tno.nl The bath-tub curve Reliability Life-Model Bayesian Networks for analysis of LED reliability 5 Source: R. Crowe, Design for Reliability, CRC Press 2001
  • 6. jan_eite.bullema@tno.nl What is a Bayesian Network? Bayesian Networks are based upon Bayes Theorem Bayesian Networks for analysis of LED reliability 6 Source: Bayesian Statistics – a brief introduction, Ken Rice, 2014 Common sense reduced to computation Pierre-Simon, marquis de Laplace (1749-1827) Inventor of Bayesian inference
  • 7. jan_eite.bullema@tno.nl What is a Bayesian Network? A Bayesian Network is a Directed Acyclic Graph Bayesian Networks for analysis of LED reliability 7 Source: TNO – report Dynamic Bayesian Network simulations of a lighting system
  • 8. jan_eite.bullema@tno.nl What is a Bayesian Network? Bayesian Network built from Expert Knowledge Step by Step Example for a simple LED system Bayesian Networks for analysis of LED reliability 8 Source: TNO – report Dynamic Bayesian Network simulations of a lighting system Step 1. System Decomposition
  • 9. jan_eite.bullema@tno.nl What is a Bayesian Network? Bayesian Network built from Expert Knowledge Step 2. Failure Mode and Effect Analysis Bayesian Networks for analysis of LED reliability 9 Source: TNO – report Dynamic Bayesian Network simulations of a lighting system
  • 10. jan_eite.bullema@tno.nl What is a Bayesian Network? Bayesian Network built from Expert Knowledge Step 3. Failure Tree from FMEA Bayesian Networks for analysis of LED reliability 10 Source: TNO – report Dynamic Bayesian Network simulations of a lighting system
  • 11. jan_eite.bullema@tno.nl What is a Bayesian Network? Bayesian Network built from Expert Knowledge Step 4. Translate Failure Tree for computation Bayesian Networks for analysis of LED reliability 11 Source: TNO Step 5. Calculate Physics of Failure A_SOL_1 <- exp(Ea_SOL/KB*(1/TE-1/T_SOL_1)) R_SOL_1 <- exp(-TIME * A_SOL_1 * (10^9/FIT_SOL)^-1) Step 6. Do a Monte Carlo Simulation
  • 12. jan_eite.bullema@tno.nl What is a Bayesian Network? Bayesian Network built from Expert Knowledge Step 7. Do Bayesian Inference Bayesian Networks for analysis of LED reliability 12 Source: TNO Step 8. Calculate Reliability Predictions
  • 13. jan_eite.bullema@tno.nl What is a Bayesian Network? Bayesian Network built from Expert Knowledge Step 9. Repeat steps 5 - 8 over a time window Bayesian Networks for analysis of LED reliability 13 Source: TNO Step 10. Analyse Reliability Predictions
  • 14. jan_eite.bullema@tno.nl Outline Bayesian Networks for Analysis of LED reliability Bayesian Networks for analysis of LED reliability 14 Overview of Reliability Science What are Bayesian Networks? Challenges in LED reliability i.e. Lifetimes > 50.000 hrs. Modern reliability approaches, e.g. MOEST, Bayesian Networks, Big Data Conclusions
  • 15. jan_eite.bullema@tno.nl Reliability Basics What is the reliability of a single LED Bayesian Networks for analysis of LED reliability 15 http://rl.omslighting.com/ledacademy/694/led-academy LED manufacturer claims that 50% of their LEDs (from any batch) will emit at least 70% of the initial lumens after 50,000 hours of operation
  • 16. jan_eite.bullema@tno.nl Reliability Basics What is the reliability of a single LED E.g. the Mean Time Between Failures Bayesian Networks for analysis of LED reliability 16 ETAP White Paper: LED Lifetime in Practice
  • 17. jan_eite.bullema@tno.nl Reliability Basics What is the reliability of a single LED e.g. Temperature Dependence of MTBF Bayesian Networks for analysis of LED reliability 17 LumiLED High Flux High Power Reliability Data
  • 18. jan_eite.bullema@tno.nl Reliability Basics What is the reliability of a single LED? e.g. Temperature Dependence of MTBF Bayesian Networks for analysis of LED reliability 18 LumiLED High Flux High Power Reliability Data
  • 19. jan_eite.bullema@tno.nl Durability of LED Systems Lifetimes > 50.000 hrs and Mission Profiles Bayesian Networks for analysis of LED reliability 19 Life time of a LED System has to be 25 000 – 50 000 hrs. Gielen et al, Development of an intelligent integrated LED system-in-package, EPMC 2011
  • 20. jan_eite.bullema@tno.nl Durability of LED Systems System behaviour is more than the sum of component behaviour Bayesian Networks for analysis of LED reliability 20 http://www.ledsmagazine.com
  • 21. jan_eite.bullema@tno.nl LED Systems are Complex Knowing the behaviour of parts is not sufficient Bayesian Networks for analysis of LED reliability 21 Bullema, Combination of Bayesian Networks and FEM models to Predict Reliability of LED Systems, ECTS 2012 Example: The ESIP Demonstrator:
  • 22. jan_eite.bullema@tno.nl LED Systems are Complex Knowing the behaviour of parts is not sufficient Bayesian Networks for analysis of LED reliability 22 Bullema, Combination of Bayesian Networks and FEM models to Predict Reliability of LED Systems, ECTS 2012 Example: The CSSL Demonstrator:
  • 23. jan_eite.bullema@tno.nl LED Systems are Complex Knowing the behaviour of parts is not sufficient Bayesian Networks for analysis of LED reliability 23 Bullema, Combination of Bayesian Networks and FEM models to Predict Reliability of LED Systems, ECTS 2012 Example: The ESIP Demonstrator: compact LED system with integrated driver and string of six LEDs LED die Driver chip Thermal Pad IO leads Silicone lens/filler Moulding compound LED die Driver chip Thermal Pad IO leads Silicone lens/filler Moulding compound
  • 24. jan_eite.bullema@tno.nl LED Systems are Complex Knowing the behaviour of parts is not sufficient Bayesian Networks for analysis of LED reliability 24 Bullema, Combination of Bayesian Networks and FEM models to Predict Reliability of LED Systems, ECTS 2012 Example: The ESIP Demonstrator:
  • 25. jan_eite.bullema@tno.nl LED Systems are Complex Knowing the behaviour of parts is not sufficient Bayesian Networks for analysis of LED reliability 25 Bullema, Combination of Bayesian Networks and FEM models to Predict Reliability of LED Systems, ECTS 2012 Example: The ESIP Demonstrator: Failure tree for electrical defects and reduced lumen output
  • 26. jan_eite.bullema@tno.nl LED Systems are Complex Knowing the behaviour of parts is not sufficient Bayesian Networks for analysis of LED reliability 26 Bullema, Combination of Bayesian Networks and FEM models to Predict Reliability of LED Systems, ECTS 2012 Example: The ESIP Demonstrator: Calculation R(t) for not cooled and heat pipe cooled (T = T not cooled - 20 C) Design ESIP with and without heat pipe
  • 27. jan_eite.bullema@tno.nl LED Systems are Complex Knowing the behaviour of parts is not sufficient Bayesian Networks for analysis of LED reliability 27 Bullema, Combination of Bayesian Networks and FEM models to Predict Reliability of LED Systems, ECTS 2012 Example: The ESIP Demonstrator: Predicted Survival Rate R(t) of LV @ 25 ºC Continuous,1xSwitch/day,10xSwitch/day Green line = 10 x switching / day Red line = 1 x switched / day Blue line = No switching
  • 28. jan_eite.bullema@tno.nl LED Systems are Complex Knowing the behaviour of parts is not sufficient Bayesian Networks for analysis of LED reliability 28 Bullema, Combination of Bayesian Networks and FEM models to Predict Reliability of LED Systems, ECTS 2012 Example: The ESIP Demonstrator: ESIP with and without heat pipe CSSL Design A, B At Low and High Temp
  • 29. jan_eite.bullema@tno.nl LED Systems are Complex Knowing the behaviour of parts is not sufficient Use Mission Profiles to Predict Reliability Bayesian Networks for analysis of LED reliability 29 Source: TNO report
  • 30. jan_eite.bullema@tno.nl Outline Bayesian Networks for Analysis of LED reliability Bayesian Networks for analysis of LED reliability 30 ------------------------- Overview of Reliability Science What are Bayesian Networks? Challenges in LED reliability i.e. Lifetimes > 50.000 hrs. Modern reliability approaches, e.g. MOEST, Bayesian Networks, Big Data Conclusions
  • 31. jan_eite.bullema@tno.nl MEOST: a jump into the future Multiple Over Stress Testing: finding the weak spot Bayesian Networks for analysis of LED reliability 31 Keki R. Bhote , World Class Reliability, ISBN 0-8144-0792-7 Various stresses and their action
  • 32. jan_eite.bullema@tno.nl The Future lies in use of Big Data Measuring all lamps real time during using advanced prognostics: all the tools are available Bayesian Networks for analysis of LED reliability 32 http://www.smartindustry.nl/
  • 33. jan_eite.bullema@tno.nl Approach to develop Reliability Model based upon a Bayesian Network Bayesian Networks for analysis of LED reliability 33 Bullema, Combination of Bayesian Networks and FEM models to Predict Reliability of LED Systems, 2012

Editor's Notes

  • #2: Reliability is a discipline that continues to increase in importance as systems become more complex, support costs increase, and defense budgets decrease. Reliability has been a recognized performance factor for at least 50 years. During World War II, the V-1 missile team, led by Dr. Wernher von Braun, developed what was probably the first reliability model. The model was based on a theory advanced by Eric Pieruschka that if the probability of survival of an element is 1/x, then the probability that a set of n identical elements will survive is (1/x)n . The formula derived from this theory is sometimes called Lusser’s law (Robert Lusser is considered a pioneer of reliability) but is more frequently known as the formula for the reliability of a series system: Rs = R1 x R2 x . . x Rn.
  • #7: Bayesian Networks are derived from Bayes Theorem - A Bayesian Network is a so-called directed acyclic graph. Basically Bayesian Networks are used to represent knowledge and the nice thing is that you can calculate with these networks and update the networks as your knowledge – based upon new insights or new measurements- increases
  • #8: In mathematics and computer science, a directed acyclic graph (, is a directed graph with no directed cycles. That is, it is formed by a collection of vertices and directed edges, each edge connecting one vertex to another, such that there is no way to start at some vertex v and follow a sequence of edges that eventually loops back to v again.[1][2][3] DAGs may be used to model many different kinds of information.
  • #9: A Bayesian Network is a so-called directed acyclic graph. Basically Bayesian Networks are used to represent knowledge and the nice thing is that you can calculate with these networks and update the networks as your knowledge – based upon new insights or new measurements- increases
  • #10: A Bayesian Network is a so-called directed acyclic graph. Basically Bayesian Networks are used to represent knowledge and the nice thing is that you can calculate with these networks and update the networks as your knowledge – based upon new insights or new measurements- increases
  • #11: A Bayesian Network is a so-called directed acyclic graph. Basically Bayesian Networks are used to represent knowledge and the nice thing is that you can calculate with these networks and update the networks as your knowledge – based upon new insights or new measurements- increases
  • #12: A Bayesian Network is a so-called directed acyclic graph. Basically Bayesian Networks are used to represent knowledge and the nice thing is that you can calculate with these networks and update the networks as your knowledge – based upon new insights or new measurements- increases
  • #13: A Bayesian Network is a so-called directed acyclic graph. Basically Bayesian Networks are used to represent knowledge and the nice thing is that you can calculate with these networks and update the networks as your knowledge – based upon new insights or new measurements- increases
  • #14: A Bayesian Network is a so-called directed acyclic graph. Basically Bayesian Networks are used to represent knowledge and the nice thing is that you can calculate with these networks and update the networks as your knowledge – based upon new insights or new measurements- increases
  • #24: It isn't just about the LED. Good LEDs can be incorporated into poorly engineered products and turn the Methuselah of lighting into the exponent of “live fast, die young.” The promise of LED lifetime is often presented in terms of hours and years but with little background data to support anything beyond vacuous promises. The statement of 100,000 hours of LED luminaire lifetime has given way to the realization that there is little consistency, very little published data, and few hard facts around so‐called luminaire lifetime numbers. The situation is better at the LED package level, where reputable manufacturers have thousands of hours of data under varying conditions. But this is not enough.
  • #34: De onbemande ruimtesonde Voyager 2 werd op 20 augustus 1977 vanaf Cape Canaveral, Florida gelanceerd aan boord van een Titan III raket. Het oorspronkelijke doel van Voyager 1 en 2 was het van dichtbij onderzoeken van Jupiter, Saturnus, de ringen van Saturnus en de grootste manen van beide planeten. Indien mogelijk - de sondes waren gebouwd om vijf jaar operationeel te zijn - zou Voyager 2 echter ook nog Uranus en Neptunus aandoen. Op 22 april 2010 ontstond er een probleem met de communicatie tussen aarde en de Voyager 2. De 33 jaar oude Voyager die zich nu op ruim 14 miljard kilometer ( 14.000.000.000 km ) bevindt stuurde signalen naar de aarde die niet te decoderen waren. De Voyager werkte naar eigen zeggen nog goed. Nasa heeft het probleem inmiddels verholpen. Makkelijk was dat niet aangezien een commando vanaf de aarde er 13 uur over doet om de Voyager te bereiken. Nadat de Voyager drie weken in "spaarstand" heeft gestaan bleek de oorzaak in het computergeheugen te zitten. Een enkel bitje was spontaan van 0 naar 1 gesprongen. Nasa heeft de computer van de sonde gereset en de sonde lijkt nu weer naar behoren te werken