Johan van den Bogaard

1
Product Lifecycle Optimization using Degradation
Models

• Johan van den Bogaard

2
Contents

• Introduction
• Boundaries with respect to maintenance
• Theory concept
• Step-by-step protocol (ROMDA)
• Case study
• Conclusions recommendations

3
Introduction background of research

• Four business drivers
• Time
• Profitability
• Functionality
• Quality / Reliability
• Next to that
• Higher product complexity
• New environmental laws and legislations

4
Introduction background of research

• Conditions to develop a method
• Economically sensible
• Environmentally sensible
• Generally applicable
• Practically this means
• Save environmental waste, save production energy
  consumption, save money, save time, save brand
  name, etc.

5
Introduction background of research

• Three design requirements
• Optimize product design in terms of reliability
  and robustness (Design)
• Provide information for re-use decisions
• Provide information for optimal preventive
  maintenance

6
Maintenance strategy options
7
Maintenance strategy
8
Introduction background of research
9
Theory concept (ROMDA)
10
Information needed for design requirements
11
Concept of Reliability Prediction and
Optimization Method
12
Reliability Prediction and Optimization Method
function
13
Reliability Prediction and Optimization Method
14
Protocol

• Determine dominant Performance Characteristic
  (output parameter responsible for dominant
  failure mechanism) and the dominant Design
  Parameters
• Use methods like Failure Mode and Effects
  Analysis, or Fault Tree Analysis
• (Qualitative methods using field information,
  engineering knowledge, knowledge of previous
  generation products)

15
Protocol

• Perform Screening Experiments
• Screening experiments provide a verification step
  for phase 1 of the protocol
• Provide insight in how noise factors influence
  the product under study (e.g. temperature,
  humidity)
• Screening experiments use testing techniques like
  DOE or Taguchi methods

16
Protocol

• Perform a degradation test
• The degradation test provides the degradation
  profiles of the design parameters and the
  performance characteristic (e.g. convex, concave,
  linear)
• Information about real degradation
• time
• Methods that could be used are
• (Acc. Degr. Tests, Step-Stress tests,
• Compressed-time tests)

17
Protocol

• Determine time-dependent functional relationship
  between PC and DPs
• The functional relationship links the DPs to the
  PC. This way stochastic design optimization is
  possible.
• For these tests the output of the degradation
  tests are used for the test setup.
• Methods that could be used in adjusted form are
  DOE and Taguchi.

18
(No Transcript)
19
Protocol

• Translate time-dependent degradation behaviour of
  the Performance Characteristic to Reliability
  Characteristics (e.g. MTTF, VTTF)
• This translation is possible when the failure
  limits are known, else these limits have to be
  determined via tests first.

20
Protocol

• Stochastic optimization of the product
• Stochastic optimization means maximizing the MTTF
  and minimizing the VTTF

21
Case study finisher module of a copier

• Protocol
• Identify dominant failure mode and identify
  dominant PC and DPs
• Perform screening experiments
• Perform a degradation test
• Determine time-dependent functional relationship
  between PC and DPs over time
• Translate time-dependent behaviour to reliability
  characteristics
• Optimize design

22
Case Study EET Project (Signature Analysis)

• Research project that is partly funded by the
  Ecology, Economy, and Technology (EET) programme
  of the Dutch Ministry of Economic Affairs (EET
  Grant EETK 20037) and partly funded by
  Flextronics.
• Partners in project
• Flextronics
• Tu/e
• Eurandom
• OCE NV
• PDE
• DTI

23
Case Study The Finisher Module
24
Case Study Finisher Module

• Finisher module of a copier machine
• Performance characteristic ? Current rise time
  (T_pr)
• (controller controls speed motor)
• Dominant parameters ? Motor Load, Tload (X1)
• (maintray experiments) (contamination,
  friction, wear)
• Resistance of PWBA, Rs (X2)
  (increase of resistance)

25
Dominant failure mechanism Paper transport
DP 1
DP 2
Mechanical load nip motor
Electric resistance PWBA
PC
Current rise time nip motor
26

• Protocol
• Identify dominant failure mode and identify
  dominant PC and DPs
• Perform screening experiments
• Perform a degradation test
• Determine time-dependent functional relationship
  between PC and DPs over time
• Translate time-dependent behaviour to reliability
  characteristics
• Optimize design

27
DP 1
DP 2
PC
28
DP 1
DP 2
29

• Protocol
• Identify dominant failure mode and identify
  dominant PC and DPs
• Perform screening experiments
• Perform a degradation test
• Determine time-dependent functional relationship
  between PC and DPs over time
• Translate time-dependent behaviour to reliability
  characteristics
• Optimize design

30
DP 1
load
PC
DP 2
RPWBA
Nr of copies
0
4,5
9
31
LSLPC 504,28 µs
32

• Protocol
• Identify dominant failure mode and identify
  dominant PC and DPs
• Perform screening experiments
• Perform a degradation test
• Determine time-dependent functional relationship
  between PC and DPs over time
• Translate time-dependent behaviour to reliability
  characteristics
• Optimize design

33
Translation to reliability char. and optimization
design
34
(No Transcript)
35
Overall conclusions

• Stochastic Optimization of Product is possible
  
• (functional relationship is known and
  translation to reliability characteristics is
  made in a stochastic way)
• Re-use of modules or sub-systems is possible
• (indicators are the PC or the DPs)
• Preventive Maintenance is possible
• (indicators are the PC or the DPs)
• ROMDA is currently being used in the Flextronics
  process for both re-use decisions and preventive
  maintenance decisions. Implementation of the
  design phase is currently initiated.
• Implementation possibilities are currently being
  researched for the German car industry.

36
Recommendations (1)

• From theoretical point of view
• Consequences of assumptions need to be researched
• Distributional assumptions (normal distributions,
  etc.)
• Research on optimal reliability characteristics
  (MTTF, SDTTF, etc.)
• Research optimal statistical modelling techniques
  (LSE, MLE, etc.)
• Research optimal degradation testing strategy
  (ADT instead of Compressed-time testing, etc.)
• Research risk reduction options for failure
  mechanism identification phase

37
Recommendations (2)

• From practical point of view
• Application domain of ROMDA with respect to
  product groups
• Translation of phase of ROMDA to Product
  Development Process for optimal implementation

38
Questions?
39
(No Transcript)
40
Theoretical Example Simulation Experiments
Temperature control system
Temperature control circuit
41
Simulation Experiments

• Transfer Function
• Input voltage E0
• Nominal voltage Zener Diode Ez
• Four resistors R1, R2, R3 and R4
• Resistor RT (Performance Char.)

42
Simulation Experiments
Input parameters Degradation path of
resistors with ? depending on
chosen degradation profile.
PC RT(t) LSL2.5 k? and USL2.9 k?
43
Reliability prediction and improvement

• Optimize models
• Combined Multiple Response Optimization
• Maximize ? of lifetime
• Minimize ? of lifetime
• Use Desirability Approach

44
Results Simulations
45
Results Simulations
46
Conclusions Simulations

• Improvement of mean lifetime of 57 percent.
• Slight improvement of standard deviation of
  lifetime.