What is Resilience Engineering? Changing focus of system safety Kazuo Furuta

What is Resilience Engineering?
Resilience Engineering Research Center
Kazuo Furuta
Resilience Engineering Research Center
© K. Furuta
Changing focus of system safety
Accident model
Linear model
Epidemiological model
Systemic model
(Single root cause) (Multiple latent factors) (Emergent variability)
System complexity
Era of resilience
Era of human errors
Era of technology
The issue is
hardware failures.
The issue is
individual
human
performance.
de Havilland Comet
1960
Resilience Engineering Research Center
The issue is
vulnerability
against
unanticipated
situations.
Era of
socio-technical
interactions
The issue is
Improper
socio-technical
interactions.
TMI-2
1980
Chernobyl
Challenger
WTC Tohoku earthquake
2000
© K. Furuta
Linear model
• Premise
– An accident occurs when a series of events occur in a
specific order.
• Causes
– Malfunctions and failures (root causes) of a definite
set system components (equipment or humans)
• Countermeasures
– Eliminate events or situations
that may become root
causes of an accident
Resilience Engineering Research Center
© K. Furuta
Principles behind linear model
• Decomposition principle
– Any system is composed of components and
understandable by reduction into the components.
• Bimodal state principle
– Functioning of any system can be described just by
two states: normal (success) or abnormal (failure).
• Independence (linear) principle
– Interactions between system components are definite,
and it can be assumed that their functions are
independent each other.
Resilience Engineering Research Center
© K. Furuta
Example of linear model
Initiator Reactivity
control Pressure boundary
Turbine
trip
TT
Reactor RV open RV close
trip
C
M
Core cooling
RHR
Feed
water
HPCI
Depressure
LPCI
Q
U
X
V
P
Sequence
W
OK
OK
TT QW
OK
TT QUW
TT QUV
TT QUX
OK
TT PW
OK
TT PQW
OK
TT PQUW
TT PQUV
TT M
TT C
Resilience Engineering Research Center
© K. Furuta
Epidemiological model
• Premise
– A combination of multiple failures and latent
conditions cause an accident.
• Causes
– Degradation of safety barriers (physical, functional,
symbolic, conceptual)
• Countermeasures
– Detect and repair
degradation of safety
barriers
organizationally
スイスチーズ・モデル
Resilience Engineering Research Center
© K. Furuta
Epidemiological view of human errors
Private factors
Environmental factors
Societal factors
Cognitive
mechanism
Context
HEP = P(Unsafe act|EFC) x P(EFC)
• Error Forcing Context (EFC)
Context such that humans inevitably make an error
Resilience Engineering Research Center
© K. Furuta
Normal accidents
• Charles Perrow (1984)
INTERACTIONS
Tight
Linear
Dams
Complex
Nuclear plants
Power grids
DNA
Chemical plants
Rail transport
COUPLING
Space mission
Airways
Military adventures
Assembly-line
Loose
Mining
R&D firms
Manufacturing
Universities
Resilience Engineering Research Center
• Accidents in a huge and
complex system are inevitable
– Unforeseen strong connections
between separate parts of system
– Non-linear system behavior
– Complexity beyond human
understanding
– Erroneous safety protections
• Acceptance of technologies
should not be determined by
risk but by potential danger.
© K. Furuta
Complex system
Chaos
Emergent phenomena
Network system
Stylized fact
Fractal
Resilience Engineering Research Center
© K. Furuta
Non-linear system and resonance
+
• Stochastic resonance
=
Tacoma Narrows, 1940
– A weak periodic signal added
to a non-linear system will
emerge in the presence of a
stochastic noise.
Resilience Engineering Research Center
© K. Furuta
Systemic model
• Premise
– The performance of system function fluctuates
continuously.
• Causes
– An unexpected combination (resonance) of
performance variability causes an accident.
• Countermeasures
– Monitoring and damping
performance variability
Resilience Engineering Research Center
© K. Furuta
Four principles of FRAM
Functional Resonance Accident Model
• Equivalence of success and failure
– Success and failure derive from essential variability of
system function. There are no difference between them.
• Approximate adjustment
– Adjustment for damping performance variability must be
incomplete and approximate due to resource limitations.
• Emergence
– System behavior emerges from variability of non-linear
system functions.
• Functional resonance
– An accident occurs when variability of system function
exceeds a safety limit due to functional resonance.
Resilience Engineering Research Center
© K. Furuta
Functional Resonance Accident Model
Function
Variability of system function
Variability of
component function
Safety boundary
Time
Resilience Engineering Research Center
© K. Furuta
What is resilience?
• Ecological resilience (Holling, 1973)
– A measure of the persistence of systems and of their
ability to absorb changes and disturbances and still
maintain the same relationships between populations
or state variables
• Seismic resilience (Bruneau, 2003)
– The ability of both physical and social systems to
withstand earthquake-generated forces and demands
and to cope with earthquake impacts through
situation assessment, rapid response, and effective
recovery strategies
Resilience Engineering Research Center
© K. Furuta
What is resilience?
• Economic resilience
– The ability to escape from serious economic crises, or
to recover from crises by mitigating the influence of
external shocks
• Business resilience
– The ability to respond and adapt quickly to internal or
external disturbances of business opportunities,
demands, confusions, and threats, to suppress their
impacts, and to continue business operations
Resilience Engineering Research Center
© K. Furuta
Resilience engineering
• Resilience from systemic view
– The intrinsic ability of a system to adjust its
functioning prior to, during, or following changes and
disturbances, so that it can sustain required
operations under both expected and unexpected
conditions.
• Resilience engineering
– Design methodology of resilience
– The objective of risk management is not reduction of
risks, but enhancement of the ability to suppress
system performance variability under changes,
disturbances, and uncertainties.
Proactive risk management
Resilience Engineering Research Center
© K. Furuta
High Reliability Organization (HRO)
• Organization where accident occurring is
suppressed below the standard level
under severe conditions
– Aircraft carrier, Air Traffic Control, Nuclear
power plants, Emergency Rescue Center
• Vigorously studied around 1990 at UCB
• Features common in HRO
– Mindfulness
– Ability to manage unanticipated situations
Resilience Engineering Research Center
© K. Furuta
Implementation process
• Anticipation
– Get ready for long-term threats and changes
• Monitoring
– Watch system states and find out clues of threats
• Responding
– Take immediate actions to regulate function
variability
• Learning
– Learn from good as well as bad consequences
Resilience Engineering Research Center
© K. Furuta
Functionality
Resilience triangle (Bruneau,2003)
Bruneau,2003)
Crisis
Damage
High resilience
Resilience
triangle
Low resilience
Time
Resilience Engineering Research Center
© K. Furuta
Assessment of resilience
• Key issue in assessing system resilience
– Different people have different interests,
valuations, needs, and so on; resilience is
different for different stakeholders.
• Demonstration with recovery of lifeline
after Tohoku Earthquake
– Maslow’s five-layered hierarchy of human
needs
– Persona method
Resilience Engineering Research Center
© K. Furuta
Maslow’s hierarchy of human needs
• Physiological
– Freedom from hunger, thirst, sleepiness, fatigue, cold, etc.
• Safety
– Protection against hazards, threats, fear, and uncertainty.
• Social
– Desire to be liked by others, to belong to community, etc.
• Esteem
– Desire to be accepted, respected, and valued by others.
• Self-actualization
– Desire to become more and more what one is.
Resilience Engineering Research Center
© K. Furuta
Decomposition of assessment measure
Level
Item
Basic data
Physiological
Water
Food
Dwelling
Medical care
Water supply, water wagons
Shops, distribution
Home, refugee camps
Hospitals
Safety
Electricity
Water
Gas
Information
Electricity grid, generators
Water supply
Gas lines
Internet, TVs, radios
Social
Privacy
Job
Relatives
Property
Home or refugee camp
Workplace, employer
State of relatives
House, cars
Resilience Engineering Research Center
© K. Furuta
Assessment for different stakeholders
• Persona
– Imaginary but very
specific user model to be
considered in designing
products or services
– Not an average user
• Characteristic three
personas of earthquake
sufferers
– Based on opened notes
of sufferers
【Persona B】
Age: 40s
Sex: male
Residence: Kesen-numa
Family: Wife and 2 sons
Health condition:
Good but blood
pressure is a little
high
Occupation:
Self-owned shop job
Resilience Engineering Research Center
© K. Furuta
Satisfaction (%)
Resilience triangle of utilities after
Tohoku Earthquake (2)
100
Social needs
0
3/8
Persona A
Persona B
Persona C
3/22
4/5
4/19
5/3
5/17
5/31
Date
Resilience Engineering Research Center
© K. Furuta
Satisfaction (%)
Resilience triangle of utilities after
Tohoku Earthquake (1)
100
Physiological needs
0
3/8
Persona A
Persona B
Persona C
3/11
3/14
3/17
3/20
3/23
Date
Resilience Engineering Research Center
© K. Furuta
Essential characteristics of resilience
Safety boundary
Margin
How closely operating
to a boundary
Buffering capacity
Size of disruptions
that system can absorb
Adaptation
Flexibility
Ability to restructure itself
in response to changes
Resilience Engineering Research Center
Functionality
(Resilience triangle)
Boundary
Tolerance
System behavior
near a boundary
(Graceful degradation)
© K. Furuta
Premises in engineering design
• Systems are to be designed to satisfy specific
design bases that are predefined.
• Situations beyond the design bases are not
assumed. Functions beyond the design bases
are not guaranteed.
• The probability that the system goes beyond the
design bases is empirically predictable.
• Where to locate the design bases will not be
determined by engineering but by economics
and politics.
Resilience Engineering Research Center
© K. Furuta
Risk retention
Probability of loss
Safety based on probabilistic risk
Risk reduction
Risk limit
Risk transfer / retention
Scale of loss
Resilience Engineering Research Center
© K. Furuta
Overview of resilience engineering
Synthesis
Assessment
Socialization
Physics of failure
Maitenology
Intelligent training
Condition monitoring
Lifetime prediction
Decision-making
Ontology
Implementation process
Anticipate
Simulation
Respond
Adapt
Incident analysis
Replanning
Human modeling
Risk discovery
Monitor
Nondestructive inspection
Performance indicator
Data mining
Energy strategy
Assessment of safety culture
Maintenance data base
Base technologies
Resilience Engineering Research Center
© K. Furuta
Super resilience
• Resilience is not restricted to system’s response
to crisis, but also includes adaptation to slow
and long term changes.
– Resolution of the 3E trilemma (sustainability)
– Endless innovation under changing global business
environment
• Society can get stronger and smarter than the
pre-event level by restructuring itself from
experience.
Super resilience
Resilience Engineering Research Center
© K. Furuta
Summary
• The conventional approach for risk
management does not work often in
actual situations. New approaches based
on a systemic viewpoint are desired.
• Resilience engineering is a promising idea
that can give solution to the above
problem based on a concept of complex
systems or the systemic model of accident.
Resilience Engineering Research Center
© K. Furuta