12004 MAPLD/Mishaps SeminarChandler Using Root Cause Analysis To Understand Failures & Accidents...

12004 MAPLD/Mishaps Seminar Chandler

Using Root Cause Analysis To Understand Failures & Accidents

Faith ChandlerOffice of Safety & Mission Assurance

7th Military and Aerospace Programmable Logic Devices (MAPLD)

September 8, 2004


Agenda

• What’s a NASA Mishap?

• Types of Mishaps

• Purpose of Mishap Investigation

• Investigating Causes of Mishaps and Failures

• Definitions

• Steps in Root Cause Analysis

• Generating Recommendations

• For More InformationStop and ask yourself…

Did you really find the causes of the failure or accident?


What’s A Mishap?

NASA Mishap. An unplanned event that results in at least one of the following:

• Injury to non-NASA personnel, caused by NASA operations.

• Damage to public or private property (including foreign property), caused by NASA operations or NASA-funded development or research projects.

• Occupational injury or occupational illness to NASA personnel.

• NASA mission failure.

• Destruction of, or damage to, NASA property.


“What can go wrong?”• Equipment will fail• Software will contain errors• Humans will make mistakes• Humans will deviate from accepted

policy and practices

There is a lot at stake!• Human life• One-of-a-kind hardware• Government equipment & facilities• Scientific knowledge• Public confidence

Types of Mishaps

Mars ClimateOrbiter

Challenger

Columbia

NOAA N Prime

HeliosPayload Canister


Purpose of NASA Mishap Investigation

• The purpose of NASA mishap investigation process is solely to determine cause and develop recommendations to prevent recurrence.

• This purpose is completely distinct from any proceedings the agency may undertake to determine civil, criminal, or administrative culpability or liability, including those that can be used to support the need for disciplinary action.


Investigating Causes of Failures & Mishaps

NavigationEquipment Failure

NavigationEquipment Failure

Circuit ShortedCircuit Shorted

SatelliteSystem Failure

SatelliteSystem Failure

Often investigators:

• Identify the part or individual that failed.

• Identify the type of failure.

• Identify the immediate cause of the failure.

• Stop the investigation.

Problem with this approach:

The underlying causes may continue to produce similar problems or mishaps in the same or related areas.

IncorrectInstallationIncorrect

Installation

Bent PinBent Pin


Like icebergs, most of the problem is usually below the surface!




When performing an investigation, it is necessary to look at more than just the immediately visible cause, which is often the proximate cause.

There are underlying organizational causes that are more difficult to see, however, they may contribute significantly to the undesired outcome and, if not corrected, they will continue to create similar types of problems. These are root causes.

Per NPR 8621.1: NASA Procedural Requirements for Mishap reporting, Investigating, and Recordkeeping, all NASA mishap and close call investigations must identify the proximate causes(s), root causes(s) and contributing factor(s).


Definitions

Proximate Cause(s) (Direct Cause)• The event(s) that occurred, including any condition(s) that existed

immediately before the undesired outcome, directly resulted in its occurrence and, if eliminated or modified, would have prevented the undesired outcome.

• Examples of proximate causes:

Equipment Human• Arched • Pushed incorrect button• Leaked • Fell• Over-loaded • Dropped tool• Over-heated • Connected wires


Root Cause(s)• One of multiple factors (events, conditions or organizational factors)

that contributed to or created the proximate cause and subsequent undesired outcome and, if eliminated, or modified would have prevented the undesired outcome. Typically multiple root causes contribute to an undesired outcome.

Organizational factors • Any operational or management structural entity that exerts control

over the system at any stage in its life cycle, including but not limited to the system’s concept development, design, fabrication, test, maintenance, operation, and disposal.

• Examples: resource management (budget, staff, training); policy (content, implementation, verification); and management decisions.

Definitions


Definitions

Root Cause Analysis (RCA)

• A structured evaluation method that identifies the root causes for an undesired outcome and the actions adequate to prevent recurrence. Root cause analysis should continue until organizational factors have been identified, or until data are exhausted.

• RCA is a method that helps professionals determine:

• What happened.

• How it happened.

• Why it happened.

• Allows learning from past problems, failures, and accidents.


Root Cause Analysis - Steps

1. Identify and clearly define the undesired outcome.

2. Gather data.

3. Create a timeline.

4. Place events & conditions on an event and causal factor tree.

5. Use a fault tree or other method/tool to identify all potential causes.

6. Decompose system failures down to a basic events or conditions (Further

describe what happened)

7. Identify specific failure modes (Immediate Causes)

8. Continue asking “WHY” to identify root causes.

9. Check your logic and your facts. Eliminate items that are not causes or

contributing factors.

10. Generate solutions that address both proximate causes and root causes.



Clearly define the undesirable outcome. • Describe the undesired outcome. • For example: “satellite failed to deploy,” “employee broke his

arm,” or “XYZ project schedule significantly slipped.”

Gather data.Identify facts surrounding the undesired outcome.

• When did the undesired outcome occur?• Where did it occur?• What conditions were present prior to its occurrence?• What controls or barriers could have prevented its occurrence

but did not? • What are all the potential causes?• What actions can prevent recurrence?• What amelioration occurred? Did it prevent further damage or

injury?



Create a timeline (sequence diagram)• Illustrate the sequence of events in chronological order

horizontally across the page.

• Depict relationships between conditions, events, and exceeded or failed barriers/controls.

Exceeded- Failed Barrier

Or Control


Or Control

EventEventUndesiredOutcome

Condition Condition

EventEventEventEvent


Root Cause Analysis - StepsCreate a timeline (sequence diagram)

• If amelioration occurred (e.g., put out the fire, cared for the injured), this should be included in the evaluation to ensure that it did not contribute to the undesired outcome.

Example: In the case of a death, the investigation should ensure that the death was the result of the mishap and not a delay in medical care or inappropriate medical care at the scene.


Or Control


Or Control

EventEventUndesiredOutcome

Condition Condition

EventEventEventEventExceeded-

FailedAmelioration

Exceeded- Failed

Amelioration



Example: simple timeline.

Satellite Failed to Deploy Antenna

Satellite Failed to Deploy Antenna

Thrusters OrientedSpace Craft

Thrusters OrientedSpace Craft

Lost HighSpeed Data

Stream(Mission Failure)

Tech. UsedWrong Method

To Correct

Tech. UsedWrong Method

To Correct

SatellitePowered UpSatellite

Powered Up

Poor Line of

Sight

Poor Line of

Sight



Create an event and causal factor tree.

(A visual representation of the causes that led to the failure or mishap.)

• Place the undesired outcome at the top of the tree.

• Add all events, conditions, and exceeded/failed barriers that occurred immediately before the undesired outcome and might have caused it.

Satellite FailedTo Deploy Antenna


Thrusters Oriented

Space Craft

Thrusters Oriented

Space Craft

Technician Used Wrong

Method to Correct

Technician Used Wrong

Method to Correct

Lost High Speed Data Stream From Satellite (Mission Failure)


SatellitePower UpSatellite

Power UpPoor

Line of SightPoor

Line of Sight



Create an event and causal factor tree.

• Brainstorm to ensure that all possible causes are included, NOT just those that you are sure are involved.

• Be sure to consider people, hardware, software, policy, procedures, and the environment.

MeteoroidImpacted Satellite



Thrusters Oriented Space Craft


Technician Used WrongMethod to Correct





Power UpPoor

Line of SightPoor

Line of Sight

Logic Control Failed

Power Supply Failed

Arming RelayFailed

Firing RelayFailed

Sabotage



Create an event and causal factor tree continued...

• If you have solid data indicating that one of the possible causes is not applicable, it can be eliminated from the tree.

Caution: Do not be too eager to eliminate early on. If there is a possibility that it is a causal factor, leave it and eliminate it later when more information is available.











Power UpPoor

Line of SightPoor

Line of Sight


Power Supply Failed

Arming RelayFailed

Firing RelayFailed

Sabotage

X


Root Cause Analysis - StepsCreate an event and causal factor tree continued…

• You may use a fault tree to determine all potential causes and to decompose the failure down to the “basic event” (e.g., system component level).








PoorLine of Sight

PoorLine of Sight


Power Supply Failed

Arming RelayFailed

Firing RelayFailed

Sabotage

Converter Failed

Battery Failed Current Detection Circuit Failed





• A fault tree can also be used to identify all possible types of human failures.

Didn’t PerceiveSystem Feedback







PoorLine of Sight

PoorLine of Sight

Didn’t Understand System Feedback



Correct InterpretationIncorrect Decision

Correct Decision ButIncorrect Action

Perception Error Interpretation Error Decision-Making Error Action-Execution Error

Rule-BasedError

Knowledge-BasedError

Skill-BasedError



• After you have identified all the possible causes, ask yourself “WHY” each may have occurred.

• Be sure to keep your questions focused on the original issue. For example “Why was the condition present?”; “Why did the event occur?”; “Why was the barrier exceeded?” or “Why did the barrier fail?”

Event #2Event #2 Failed or Exceeded Barrier or Control

Failed or Exceeded Barrier or Control

Undesired OutcomeUndesired Outcome

ConditionConditionEvent #1Event #1

WHY Event #1 Occurred








WHY ConditionExisted orChanged










WHYFailed

Exceeded Barrier or Control

WHYFailed


WHYFailed


WHYFailed


WHYFailed


WHYFailed



Root Cause Analysis – Steps

Continue to ask “why” until you have reached:

1. Root cause(s) - including all organizational factors that exert control over the design, fabrication, development, maintenance, operation, and disposal of the system.

2. A problem that is not correctable by NASA or NASA contractor.

3. Insufficient data to continue.


Root Cause Analysis- Steps

The resultant tree of questions and answers should lead to a comprehensive picture of POTENTIAL causes for the undesired outcome








WHY WHY WHYWHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY

WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY













WHYFailed


WHYFailed


WHYFailed


WHYFailed


XWHYFailed


WHYFailed



WHYFailed


WHYFailed






WHYFailed


WHYFailed










Check your logic with a detailed review of each potential cause. • Verify it is a contributor or cause.• If the action, deficiency, or decision in question were corrected, eliminated

or avoided, would the undesired outcome be prevented or avoided? > If no, then eliminate it from the tree.









WHY WHY WHYWHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY

WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY

XX X

XXX X X XX X

X X X X X X X

XX WHYFailed


WHYFailed



Create an event and causal factor tree continued…• The remaining items on the tree are the causes (or probable causes). necessary to

produce the undesired outcome.• Proximate causes are those immediately before the undesired outcome.• Intermediate causes are those between the proximate and root causes.• Root causes are organizational factors or systemic problems located at the bottom of the tree.


ROOT CAUSES

PROXIMATE CAUSES

INTERMEDIATE CAUSES







WHYFailed/Exceeded Barrier or Control



WHYWHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY

WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY

WHY ConditionExisted or Changed




Some people choose to leave contributing factors on the tree to showall factors that influenced the event.

Contributing factor: An event or condition that may have contributed to the occurrence of an undesired outcome but, if eliminated or modified, would not by itself have prevented the occurrence.

If this is done, illustrate them differently (e.g., dotted line boxes and arrows) so that it is clear that they are not causes.


Contributing Factors

Event #2 Failed or Exceeded Barrier or Control









WHYWHY WHY WHY WHY WHY WHY WHY WHY WHY WHY WHY

WHY WHY WHY WHY WHY WHY WHY




WHY WHY WHY


InstalledImproperly

Beyond ShelfLimit

Battery Dead

Power Supply Failed

Root Cause is Much DeeperKeep Asking Why








PoorLine of Sight

PoorLine of Sight




MMOD HitSpace Craft

After Oriented

InstalledImproperly

Beyond ShelfLimit

Battery Failed

Power Supply Failed








PoorLine of Sight

PoorLine of Sight



No Quality Inspection

Insufficient Quality Staff

Insufficient Budget

ProcedureIncorrect

Not Updated

New Task InsufficientAnomaly Training

Training Does Not Exist

Not Under Configuration Mgmt

Insufficient Training Budget

Organization Under Estimates Importance of

Anomaly Training

Correct InterpretationIncorrect Decision

Decision-Making Error


Generating Recommendations:

At a minimum corrective actions should be generated to eliminate proximate causes and eliminate or mitigate the negative effects of root causes.

When multiple causes exist, there is limited budget, or it is difficult to determine what should be corrected:

• Quantitative analysis can be used to determine the total contribution of each cause to the undesirable outcome (see NASA Fault Tree Handbook, Version 1.1, for more information).

• Fishbone diagrams (or other methods) can be used to arrange causes in order of their importance.

• Those causes which contribute most to the undesirable outcome should be eliminated or the negative effects should be mitigated to minimize risk.



For More Information

• NASA PBMA Mishap Investigation Website

(http://ai-pbma-kms.intranets.com/login.asp?link=)– Includes:

• Links (e.g, to Root Cause Analysis Software, a RCA Library).

• Documents (e.g., Methods, Techniques, Tools, Publications and Presentations).

• Threaded Discussions and Polls.

• HQ Office of Safety & Mission Assurance– [email protected]


BACK-UP SLIDES


Cause (Causal Factor) An event or condition that results in an effect. Anything that shapes or influences the outcome.

Proximate Cause(s) The event(s) that occurred, including any condition(s) that existed immediately before the undesired outcome, directly resulted in its occurrence and, if eliminated or modified, would have prevented the undesired outcome. Also known as the direct cause(s).

Root Cause(s) One of multiple factors (events, conditions or organizational factors) that contributed to or created the proximate cause and subsequent undesired outcome and, if eliminated, or modified would have prevented the undesired outcome. Typically multiple root causes contribute to an undesired outcome.

Root Cause Analysis (RCA) A structured evaluation method that identifies the root causes for an undesired outcome and the actions adequate to prevent recurrence. Root cause analysis should continue until organizational factors have been identified, or until data are exhausted.

Event A real-time occurrence describing one discrete action, typically an error, failure, or malfunction. Examples: pipe broke, power lost, lightning struck, person opened valve, etc…

Condition Any as-found state, whether or not resulting from an event, that may have safety, health, quality, security, operational, or environmental implications.

Organizational Factors Any operational or management structural entity that exerts control over the system at any stage in its life cycle, including but not limited to the system’s concept development, design, fabrication, test, maintenance, operation, and disposal.

Examples: resource management (budget, staff, training); policy (content, implementation, verification); and management decisions.

Contributing Factor An event or condition that may have contributed to the occurrence of an undesired outcome but, if eliminated or modified, would not by itself have prevented the occurrence.

Barrier A physical device or an administrative control used to reduce risk of the undesired outcome to an acceptable level. Barriers can provide physical intervention (e.g., a guardrail) or procedural separation in time and space (e.g., lock-out-tag-out procedure).

Definitions of RCA & Related Terms


Classification Level Property Damage Injury

Type A Total direct cost of mission failure and property damage is $1,000,000 or more,

or

Crewed aircraft hull loss has occurred,

or

Occurrence of an unexpected aircraft departure from controlled flight (except high performance jet/test aircraft such as F-15, F-16, F/A-18, T-38, and T-34, when engaged in flight test activities).

Occupational injury and/or illness that resulted in:

A fatality,

or

A permanent total disability,

or

The hospitalization for inpatient care of 3 or more people within 30 workdays of the mishap.

Type B Total direct cost of mission failure and property damage of at least $250,000 but less than $1,000,000.

Occupational injury and/or illness has resulted in permanent partial disability.

or

The hospitalization for inpatient care of 1-2 people within 30 workdays of the mishap.

Type C Total direct cost of mission failure and property damage of at least $25,000 but less than $250,000.

Nonfatal occupational injury or illness that caused any workdays away from work, restricted duty, or transfer to another job beyond the workday or shift on which it occurred.

Type D Total direct cost of mission failure and property damage of at least $1,000 but less than $25,000.

Any nonfatal OSHA recordable occupational injury and/or illness that does not meet the definition of a Type C mishap.

Close Call Total direct cost of mission failure and property damage is less than $1,000

or

An occurrence or condition of employee concern in which there is no property damage but possesses the potential to cause a mishap.

Minor injury requiring first aid which possesses the potential to cause a mishap

or

An occurrence or condition with no injury but possesses the potential to cause a mishap.

NASA Mishap Classification Levels

12004 MAPLD/Mishaps SeminarChandler Using Root Cause Analysis To Understand Failures & Accidents...

Documents

Transcript of 12004 MAPLD/Mishaps SeminarChandler Using Root Cause Analysis To Understand Failures & Accidents...