The Powers of Fault Tree

Analysis
Bill Vesely
Office of Safety and Mission Assurance
NASA Headquarters

Fault Tree Analysis: a

Systematic and Stylized
Deductive Process
An undesired event is defined
The event is resolved into its immediate causes
This resolution of events continues until basic
causes are identified
A logical diagram called a fault tree is
constructed in the process of carrying out the
analysis

Why Fault Tree Analysis (FTA) is

carried out
To
To
To
To
To
To
To
To
To

gain an understanding of the system

document the failure relationships of the system
exhaustively identify the causes of a failure
assure compliance with requirements or a goal
identify any weaknesses in a system
prioritize contributors to failure
identify effective upgrades to a system
optimize operations and processing
quantify the failure probability and contributors

The Fault Tree

FTA produces a Fault Tree.
The fault tree is the logical model of the
relationship of the undesired event to more basic
events.
The top event of the fault tree is the undesired
event.
The middle events are intermediate events.
The bottom of the fault tree is the causal basic
events or primary events.
The logical relationships of the events are shown
by logical symbols or gates.
4

Basic Fault Tree Structure

Top Undesired
Event
Logic Gates
Intermediate
Events

Basic Events

OVERRUN OF ANY MOTOR

AFTER TEST IS
INITIATIED
G019

EMF APPLIED TO MOTOR

2 FOR t>60 SEC

EMF APPLIED TO MOTOR

1 FOR t>60 SEC

G020

EMF APPLIED TO MOTOR

3 FOR T>60 SEC

G021

G022

KS RELAY CONTACTS
REMAIN CLOSED FOR
T>60 SEC

K2 RELAY CONTACT
FAILS TO OPEN WHEN K5
RELAY CONTACTS CLOSED
FOR T>60 SEC

G023

G024

A Typical

EMF REMAINS ON K5
COIL FOR T>60 SEC

KS RELAY CONTACTS
FAIL TO OPEN

K2 RELAY CONTACTS
FAIL TO OPEN

EMF NOT REMOVED FROM

K2 RELAY COIL WHEN K5
CONTACTS CLOSED FOR
t>60 SEC

G025

B026

B028

G030

Fault Tree

K3 RELAY CONTACTS
REMAIN CLOSED FOR
T>60 SEC

K1 RELAY CONTACTS
FAIL TO OPEN WHEN K3
CONTACTS CLOSED FOR
t>60 SEC

K1 RELAY CONTACTS
FAIL TO OPEN WHEN K5
CONTACTS CLOSED FOR
t>60 SEC

EMF TO K2 COIL THRU

S1, KT1, KT2 AND KT3
CONTACTS

G027

G028

G031

B032

TEST SIGNAL REMAINS

ON K3 COIL FOR t>60
SEC

K5 RELAY CONTACTS
FAIL TO OPEN

EMF TO K1 COIL THRU

TIMER CIRCUIT WHEN K5
CONTACTS CLOSED FOR
t>60 SEC

B042

B043

G097

EMF TO K1 COIL THRU

S1 CONTACTS WHEN K5
CONTACTS CLOSED FOR
t>60 SEC
G098

EMF NOT REMOVED FROM

K1 RELAY COIL WHEN K5
CONTACTS CLOSED FOR
t>60 SEC

S1 SWITCH
INASDVERTENTLY CLOSES
OR FAILS TO OPEN

RESET SIGNAL
INADVERTENTLY APPLIED
OR NOT REMOVED FROM
SWITCH S1

G048

B100

B101

KT1 TIMER RESET

KT2 TIMER CONTACTS

FAIL TO OPEN WHEN K5
CONTACTS CLOSED FOR
t>60 SEC

KT3 TIMER RESET

B050

G090

B075

KT2 TIMER CONTACTS

FAIL TO OPEN

KT2 TIMER DOES NOT

"TIME OUT" DUE TO
IMPROPER INSTALLATION
OR SETTING

B095

B096

Applications of FTA
Prioritization of Contributors for Resource
Allocation
Development of a Design
Determination of Effective Tradeoffs
Resolution of Causes for Mishap Analysis
Demonstration of Compliance with Single Failure
Criteria
Establishment of Contingency Criteria
Monitoring and Tracking of Performance

The Power of FTA in Prioritizing

Failure Contributors
Each basic event in the fault tree can be prioritized for its
importance to the top event
Different importance measures are obtained for different
applications
Basic events generally are ordered by orders of magnitude
in their importance.
In addition to each basic event, every intermediate event
in the FT can be prioritized for its importance
As a general rule, less than 20% of the contributors result
in more than 90% of the risk.

Basic Fault Tree Importance

Measures

FV Importance = Relative contribution to the system

failure
probability from a component failure
RAW = Factor increase in the system failure probability
when a
component is assumed to be failed
RRW = Factor decrease in the system failure probability
when a
component is assumed to succeed
FV Importance = Fussell-Vesely Importance
RAW = Risk Achievement Worth
RRW = Risk Reduction Worth

Basic Causal Importances for a

Monopropellant System
FV Importance
(Contribution)

RRW Factor
(Reduction)

99.3%

143

100

86.7%

7.5

43

Relay K6 Fail to Open

13%

1.15

43

Switch S3 Fail to Open

0.5%

1.01

100

Isolation Valve IV2 Fail to

Close

0.3%

1.00

13

0.3%

1.00

1.00

0.01%

1.00

1.00

Basic Causal Event

Human Error Failure to
Open Switch S3
Timer K6 Fail to Time Out

Relay K3 Fail to Open

Isolation Valve IV3 Fail to
Close

10

RAW Factor
(Increase)

Uses of the Importance

Measures
Focus system safety on the top contributors
(FV)
Review possible relaxations for the lowest
contributors (FV, RAW)
Focus on upgrades having the greatest
improvements (RRW)
Define contingency measures to be consistent
with the failure impact (RAW)
Establish assurance requirements to be
consistent with their importance (FV, RAW)
11

Examples of Importance Evaluations in the

Space Shuttle PRA
Over a million individual events are modeled in the Shuttle
PRA and 97% of the calculated risk resides in
approximately 308 events.
Approximately 15% or more of the calculated risk is due to
fluid leaks that lead to fire and explosion. This can change
based on current updating of the Shuttle PRA
Abort risk is insignificant to mission risk (<1%).

12

The Use of FTA in Mishap

Analysis
The accident scenario is constructed for the
mishap
System failures (pivotal events) are identified
which resulted in the mishap
A fault tree is constructed for each system
failure to resolve the basic events involved
Root cause analysis is carried out by further
resolving a basic event into its root causes
The basic events and root causes are
dispositioned into their importances and
actions required
13

The
Columbia
Fault Tree

LOCV DURING
ENTRY DUE TO
AERODYNAMIC BREAKUP

LOCV
AERODYNAMIC
BREAKUP DUE TO
IMPROPER ATTITUDE
/ TRAJECTORY
CONTROL

AERODYNAMIC
BREAKUP DUE TO
STRUCTURAL FAILURE
OF THE ORBITER

36
AC

SF

STRUCTURAL FAILURE
OF ORBITER DUE
TO LOSS OF STRUCTURAL
MEMBER

LOSS OF AERODYNAMIC
CHARACTERISTICS
DUE TO LOSS OF
OUTER MOLD LINE (OML)

34
SFSM

LOSS OF OML
DUE TO WING/ELEVON
STRUCTURAL FAILURE

LOSS OF OML
DUE TO AFT FUSELAGE
FAILURE

SFOML

LOSS OF OML
DUE TO PAYLOAD
BAY DOOR FAILURE

LOSS OF OML
DUE TO CREW CABIN
MODULE FAILURE

LOSS OF OML
DUE TO BODY FLAP
FAILURE

LOSS OF OML
DUE TO FRCS STRUCTURAL
FAILURE

2
SFOML-WING

LOSS OF OML
DUE TO FWD FUSELAGE
FAILURE

SFOML-FWD

SFOML-AFT

LOSS OF OML
DUE TO MID FUSELAGE
FAILURE

SFOML-MID

SFOML-BAY

SFOML-CABIN

LOSS OF OML
DUE TO OMS POD
FAILURE

LOSS OF OML
DUE TO TAIL FAILURE

SFOML-OMS

SFOML-TAIL

SFOML-FLAP

SFOML-FRCS

LOSS OF OML
DUE TO WINDOW
FAILURE

LOSS OF OML
DUE TO PAILURE
SOURCE INTERNAL
TO PAYLOAD BAY

SFOML-WINDOW

SFOML-INTPLB

LOSS OF OML DUE

TO SSME OUT OF
CONFIGURATION
(ME)

SFOML-SSME

LOCV - LOCV DURING ENTRY DUE TO AERODYNAMIC BREAKUP

14

2003/04/18

Page 1

The Columbia Fault

Tree Continued

LOSS OF OML
DUE TO WING/ELEVON
STRUCTURAL FAILURE

SFOML-WING

WING/ELEVON
STRUCTURAL FAILURE
DUE TO OVERPRESSURIZATION
OR COLLAPSE

WING/ELEVON
STRUCTURAL FAILURE
DUE TO WING OVERLOAD

WING/ELEVON
STRUCTURAL FAILURE
DUE TO THERMAL
DAMAGE BURN THROUGH

21

SFOML-WING-5-1

WING OVERPRESS OR
COLLAPSE DUE TO PRESSURE
SYSTEM FAILURE CAUSING
INADVERTENT PRESS

INADVERTENT
PRESSURIZATION
DUE TO PRSD/ECLSS
TANK FAILURE

SFOML-WING-5-2

WING OVERPRESS
OR COLLAPSE DUE
TO PAYLOAD BAY
SOURCE (ME)

SFOML-WING-6-6

WING/ELEVON
STRUCTURAL FAILURE
DUE TO WEAKENED
WING STRUCTURE

SFOML-WING-6-18

23

SFOML-WING-5-3

WING OVERPRESS OR
COLLAPSE DUE TO BLOCKED
VENT RESULTING IN FAILURE
TO REPRESS

SFOML-WING-6-5

SFOML-WING-5-4

WING OVERPRESS OR
COLLAPSE DUE TO BLOWN
TIRE CAUSING INADVERTENT
PRESSURIZATION

SFOML-WING-6-7

INADVERTENT
PRESSURIZATION
DUE TO OTHER
EVENT

3
SFOML-WING-7-12

SFOML-WING-7-13

SFOML-WING - LOSS OF OML DUE TO WING/ELEVON STRUCTURAL FAILURE

15

2003/04/18

Page 2

The Columbia
Fault Tree
Continued

WING/ELEVON STRUCTURAL
FAILURE DUE TO THERMAL
DAMAGE BURN THROUGH

SFOML-WING-5-3
OVERHEAT/ THERMAL
DAMAGE/BURN THROUGH DUE
TO INADVERTENT OPENING IN
WING ALLOWING PLASMA FLOW

OVERHEAT/BURN
THROUGH DUE TO TPS
MALFUNCTION

THERMAL DAMAGE
BURN THROUGH
DUE TO HIGHER
HEATING

6
SFOML-WING-6-10

TPS MALFUNCTION
DUE TO REINFORCED
CARBON-CARBON
(RCC) FAILURE

SFOML-WING-6-9

TPS MALFUNCTION
DUE TO BLANKET
FAILURE

TPS MALFUNCTION
DUE TO TILE FAILURE

18

20

SFOML-WING-7-20

SFOML-WING-7-21

SFOML-WING-7-22

SEAL FAILURE
DUE TO THERMAL
BARRIER FAILURE

TPS MALFUNCTION
DUE TO SEAL FAILURE

INADVERTENT DOOR OPEN

OR BREACH OF THERMAL
SEAL (GEAR REMAINS UP)

SFOML-WING-7-23

SFOML-WING-7-18

SEAL FAILURE
DUE TO ENVIRONMENTAL
SEAL FAILURE

SFOML-WING-6-8

INADVERTENT OPENING IN
WING ALLOWING PLASMA
FLOW DUE TO FLIPPER
DOOR FAILURE

SFOML-WING-7-19

SEAL FAILURE
DUE TO ELEVON
COVE SEAL FAILURE

7
SFOML-WING-8-22

SFOML-WING-8-24

THERMAL BARRIER FAILURE

DUE TO LANDING GEAR
THERMAL BARRIER FAILURE

THERMAL BARRIER FAILURE

DUE TO ET UMBILICAL DOOR
THERMAL BARRIER FAILURE

SFOML-WING-9-32

SFOML-WING-9-33

SFOML-WING-8-23

IMPACT

12
SFOML-WING-10-48

SNEAK FLOW

LANDING GEAR THERMAL

BARRIER FAILURE DUE TO
FLIGHT OUTSIDE ENVELOPE
(ME)

LANDING GEAR THERMAL

BARRIER FAILURE DUE TO
INSTALL/CLOSING

LANDING GEAR THERMAL

BARRIER FAILURE DUE TO
DEGRADATION OF THERMAL
BARRIER

SFOML-WING-10-18

SFOML-WING-10-17

11
SFOML-WING-9-24

SFOML-WING-9-46

SFOML-WING-5-3 - WING/ELEVON STRUCTURAL FAILURE DUE TO THERMAL DAMAGE BURN THROUGH

16

2003/04/18

Page 4

The Columbia
Fault Tree
Continued
INFLIGHT
IMPACT ON
ASCENT

SFOML-WING-10-3

SRB DEBRIS IMPACT

ON ASCENT (ME)

ET DEBRIS IMPACTON
ASCENT (ME)

SFOML-WING-11-8

SFOML-WING-11-5

RSRM DEBRIS IMPACT

ON ASCENT (ME)

LIFT OFF (PAD

DEBRIS) IMPACT (ME)

SFOML-WING-11-7

SFOML-WING-11-18

ORBITER CAUSED
INFLIGHT IMPACT ON
ASCENT

SFOML-WING-11-6

SFOML-WING-11-64
FOD

ORBITER
HARDWARE

SFOML-WING-12-25
TPS IMPACTS
ORBITER

SFOML-WING-11-60

SFOML-WING-10-3 - INFLIGHT IMPACT ON ASCENT

SSME DEBRIS
IMPACT (ME)

SFOML-WING-12-24
ORBITER ACCESS
PANEL

SFOML-WING-11-61

17

2003/04/18

Page 15

The Use of FTA in Design

To evaluate a Design, a top level fault tree is developed
Functional level
System level
Subsystem level
Tradeoffs are carried out
Alternative functional capabilities
Alternative redundancies
Allocations are performed
System requirement into subsystem requirements
Subsystem requirements into component
requirements

18

Redundant Seal Design

Allocation Considering
Independent Failures
REDUNDANT SEALS FAIL
TO PROVIDE
CONTAINMENT
ASSURANCE
G001

1st O-RING SEAL FAILS

B001

3rd O-RING SEAL FAILS

B003
1.00E-03

1.00E-03

2nd O-RING SEAL FAILS

B002
1.00E-03

19

Redundant Seal Design Allocation

Including Common Cause Failures
REDUNDANT SEALS
FAIL TO PROVIDE
CONTAINMENT
ASSURANCE
G001
1.00E-04

COMMON CAUSE SEAL

FAILURE

SEALS FAIL
INDEP ENDENTLY

B004

G003
1.00E-04

1.00E-09

1st O-RING SEAL FAILS

3rd O-RING SEAL FAILS

B001

B003
1.00E-03

1.00E-03

2nd O-RING SEAL

FAILS

B002
1.00E-03

20

The Fault Tree as a Master

Logic Diagram
The Master Logic Diagram (MLD) is a fault tree
identifying all the hazards affecting a system or
mission
The Master Logic Diagram can also be called a
Master Hazards Diagram (MHD)
The MLD or MHD is developed using fault tree logic
The basic events of a system MHD are the hazards
that can initiate component failures or increase
their likelihood
The basic events of a mission MLD are the hazards
that are the initiating events of potential accident
scenarios
21

Extending a System Fault Tree

to a Master Hazard Diagram
(MHD)
The top event is defined as a system failure
event
The fault tree is developed to the basic
component level
Each component failure is further resolved into
hazards and conditions that can cause failure
or increase its likelihood
The resulting system MHD identifies the
hazards affecting the system and their
consequences
Of particular importance are single failures and
hazards affecting multiple redundant
22

Ranking the Criticality of

Hazards Using FTA
Each hazard is linked to a basic event or events on
the fault tree
Equivalently each hazard is linked to the basic
events in the minimal cutsets
The criticality of the hazard is the likelihood of the
hazard times the importance of the basic event
The component importance is determined from the
FTA
The likelihood is determined from the hazard
analysis
Hazard Criticality=Likelihood x Importance of
Components Affected
23

Mission

MIR HABITABILITY
*
Page 2

Critical
Functions

Tasks

Attitude Control

Pressure Control

Maintain
Pressure
Boundary

Isolate
Compartments

Emergency
Isolation

Activate & Monitor

Attitude Control

External
Skin

Internal
Hatches

Emergency
Hatches

Gyrodynes
Automatic
Control

Mir
Thrusters

Progress
Thrusters

Progress

Electric Fuel Attitude

Power
Control
(Page 4)
Computer

Fuel

Fuel

Resources

*
Support
Systems
Activated Manually

Challenged by Collision
Challenged by Manual Action
Equipment Failure

Disabled/Challenged by Human Error

24

MIR HABITABILITY
*
Page 1

Elektron
Generator

Electric
Power
(Page 4)

Page 3

Oxygen
Generation

CO2
Control

Activate and
Monitor Oxygen
Supply

Activate and
Monitor CO2
Control

Oxygen
Candles

Equipment
Cooling VGR

Oxygen
Bottles

Vozdukh

Progress

Electric
Power
(Page 4)

Lithium
Hydroxide

Vacuum
Control
Valve

25

Bypass

Other
Russian
Systems

The Mirror Success Tree (ST)

A Success Tree (ST) identifies all the ways in
which the top event cannot occur
The ST is the complement of the FT
The ST is the mirror of the FT
The ST is useful in showing the explicit ways to
prevent the occurrence of the FT
The ST is the success space twin of the FT

26

Developing the Success Tree

from the Fault Tree
Complement the top event to a NOT event
Complement all intermediate events to NOT
events
Complement all basic events to NOT events
Change all AND gates to OR gates
Change all OR gates to AND gates
The tree is now the ST
The minimal cut sets of the ST are now called
the minimal path sets
27

The Minimal Path Sets Define

the Success Modes of the
System
A minimal path set is the smallest number of
events which if they all do not occur then the
top event will not occur
If the events in one path set are prevented to
occur then the top event will be guaranteed to
not occur
The minimal path sets are the totality of ways
to prevent the top event based on the fault
tree
The minimal paths should be determined as a
part of a fault tree analysis
28

FTA Project Management Tasks

(1)
Define the FTA
Top Event
Scope
Resolution
Assemble the project Team
FT analyst
System engineering support
Data support
Software support
Define the FTA Operational Framework
Assemble the as built drawings
FT naming scheme
Interfaces/Support to be modeled
Software to be used
29

FTA Project Management Tasks

(2)
Assemble the data
Generically applicable data
Specifically applicable data
Prepare the software package
Familiarization
Test problems
Keep a log on the FTA work
Operational and design assumptions
Events not modeled and why
Success and failure definitions
Special models and quantifications used

30

FTA Project Management Tasks

(3)
Review the work at stages
FT construction
Qualitative evaluations
Quantitative evaluations
Check and validate the results
Engineering logic checks
Consistency checks with experience
Prepare and disseminate the draft report
Conclusions/findings
FTA results
FTs
Software inputs/outputs
Obtain feedback and modify and final report
Disseminate the report
Present findings
31

Reference
Fault Tree Handbook with Aerospace Applications,
Version 1.1, NASA Publication, August 2002.

32