Microsoft Root Cause Analysis Webinar Introduction V2 987 RCA By Ops A La Carte

User Manual: 987

Open the PDF directly: View PDF .
Page Count: 106 [warning: Documents this large are best viewed by clicking the View PDF Link!]

Root Cause Analysis Webinar

SPONSORS

Ops A La Carte

DfR Solutions

SigmaQuest

SPONSORS

Ops A La Carte is a Professional Consulting Firm

focused on Reliability Engineering Services,

Reliability Management, and Reliability Education to

assist you in developing and executing any and all

elements of Reliability throughout your Organization

and your Product’s Life Cycle. We work in the area

of Electronics, Mechanical Systems, and Software.

In the area of RCA, Ops A La Carte has performed

countless root-cause analyses in the area of

electronics, mechanics, and software.

SPONSORS

DfR Solutions has world-renowned expertise in applying

the science of Reliability Physics to electrical and

electronics technologies, and the company is a leading

provider of quality, reliability, and durability (QRD)

research and consulting for the electronics industry.

DfR's integrated use of Physics-of-Failure (PoF) and

Best Practices provides crucial insights and solutions

early in product design and development, and

throughout the product life cycle.

In the area of RCA, DfR Solutions has their own failure

analysis lab in Maryland and has performed over 250

root-cause investigations in the past 4 years

SPONSORS

SigmaQuest provides an on-demand suite of solutions

that help companies build better products using

business intelligence techniques for product design,

manufacturing, supplier quality, repair and returns.

Benefits are reduced warranty costs; improved

product quality, lower costs of goods sold, and

increased revenue and profits.

In the area of RCA, SigmaQuest is well positioned

because its solutions can be used for collecting

failure data for use in the critical step of analyzing

and gathering data/evidence.

Understanding the Motivation and Basics

of Root-Cause Analysis in Electronics

Abstract: Before successful Root-Cause Analysis can begin, organizations and individuals must

understand the need to have basic problem solving skills, tools and knowledge of how

problems occur and how they can be fixed.

This portion of the webinar will discuss the fundamentals of RCA and cover some of the best

practices in the electronics industry from the Physics of Failure point of view.

James McLeish, CRE

Background: Jim McLeish

Education:

Dual EE/ME MS in Electronics Control Systems

ASQ-CRE (American Society of Quality - Certified Reliability Engineer)

32 years of Automotive, Military and Industrial Electrical/Electronics (E/E)

Part 1: Product Design, Development, Systems Engineering & Production

3 Patents Electronic Control Systems

EE System Engineering and Architecture Planning

Product Engineering Management

Part 2: Validation, Reliability, Quality Assurance, Warranty Problem Solving

& Test Technology Development

Variety of Management & Technical Leadership Positions:

Part 3: Senior Technical Staff/Consulting Associate - Design for Reliability Solutions.

Principle Investigator for E/E Failure Analysis and Root Cause Problem Solving.

E/E Manufacturing Process Optimization, Yield Improvement.

Reliability Demonstration, Product Validation and Accelerated Testing.

Field Return/ Warranty Analysis

Design Reviews for Proactive Problem Prevention

Society of Automotive Engineering (SAE) - Reliability Committee

DOD MIL-HDBK-217 Update & Enhancement Tea

Background: DfR Solutions

DfR Solutions is an Engineering- Laboratory Services and Consulting firm experienced

in Physics of Failure based Quality, Reliability and Durability (QRD) research, consulting

and applied science for electrical and electronics products and technologies.

The DfR staff provide knowledge and science based solutions that maximize

product integrity and accelerate product assurance activities.

DfR captures the broad range of reliability and quality issues in electronics through the

expansive expertise of our multi-discipline staff.

Physicists, Material Scientists, Chemists and Electronic Engineers from Various Industry Segment.

Over 500 failure analysis and root-cause investigations in the past 4 years,

A world leader in failure analysis in electronics.

Strong partnerships with the leading companies in the field of electronics,

DfR strives to make our clients life easier by providing knowledge based solutions for

electronic quality, reliability and durability issues.

From component specifications and computer modeling based lifetime predictions.

From robust design of products and process to accelerated product qualification.

From technology insertion to RCA and failure analysis.

1) Motivation for Root Cause Analysis

The 1st rule of business is now:

“The competitor who Consistently, Reliably and Profitably

provides the greatest value to customers FIRST wins.”

2nd rule is:

"There are NO OTHER RULES".

In other words it’s Survival of the Fittest and the Best.

Continuously Improvement is Essential to Becoming

and Staying the Best or At Least Remaining Competitive.

1) Motivation For Root Cause Analysis

- Continuous Improvement (C.I.)

Continuously Improvement is the ongoing effort to improve products,

services or processes, in order to advance the goals of an organization,

business or society.

A never ending effort to discover and eliminate:

Inefficient process road block and bottle necks,

Non value added activities,

Problems,

Either “incremental” improvement over time or

“breakthrough” improvement all at once.

Japanese Version Kaizen

- “Change for the Better”.

Examples of C.I. Tools

Statistical Process Control

6 Sigma Quality

Best Practices / Leasons Learned

Process Optimization

Problem Solving

2) Introduction to Root Cause Analysis

-Problem Solving

Problem Solving is an integral part of cognitive thinking & decision

making. It is essential to many aspect of daily live, it involves:

Using tools to obtain relevant data, information and knowledge,

Creating mental models of situations and how the world works,

Make logical connections that lead to the formation of potential solution

concepts,

Evaluate the potential solutions against goals, constrains and desires.

Problem solving method examples:

Trial-and-error

Brainstorming

Root Cause Analysis

Problem Solving uses similar skills as:

Solving a puzzle

Detective work.

2) Problem Solving, Failure Analysis & Continuous Improvement

Has Been the Basis of Engineering Since

Humans First Make Tools & Structures

Lessons Learned for Problem Solving

During the Construction of the Early

Step & Bent Pyramid Enabled the Ancient Egyptians

To Later Build Bigger & Better Pyramids

3) Introduction to Root Cause Analysis (RCA)

Root Cause Analysis - is a category or problem solving methods that focus on

identifying the ultimate underlying reason of why an event occurred.

Based on the belief that problems are more effectively solved by correcting or

eliminating the root causes, rather than merely addressing the obvious symptoms.

The root cause is the trigger point in a causal chain of events, which may be natural

or man-made, active or passive, initiating or permitting, obvious or hidden.

Efforts to prevent or mitigate the trigger event are expected to prevent the outcome

or at least reduce the potential for problem recurrence.

RCA is a full-blown analysis that identifies the chain of physical and human related

root cause(s) behind an undesirable event .

This differs from basic troubleshooting and problem-solving processes that typically

seek solutions to specific, relatively simple difficulties.

The undesired event may be a product durability failure, a safety incident, a customer

complaint, a quality defects, human error . . . etc.

It helps focus CA/PA (Corrective Action / Preventive Action) efforts at the points of

most leverage it is essential for pointing change management efforts in the right

direction.

3) Introduction to Root Cause Analysis

- Failure Analysis (FA)

Failure Analysis is a subcategory of RCA techniques

Systematic examination of “Failed Devices”

to determine the root cause of failure.

Use knowledge gained to improve

technology, quality and reliability.

Primarily associated with the physics and material science of mechanical, structural

and E/E (Electrical /Electronic) devices and materials (i.e. hardware).

Software FA is a growing subcategory involving computer science & programming.

Forensic Engineering a subcategory that uses science and technology to investigate

materials, structures, products or components that fail or malfunctions to establish

facts for criminal or civil legal actions.

3) Introduction to Root Cause Analysis

- Failure Analysis (FA)

Failure analysis is designed to:

Identify the failure modes (the way the product failed),

Identify the failure site (where in the product failure occurred),

Identify the failure mechanism

(the physical phenomena involved in the failure),

Determine the root cause

(the design, defect, or loads which led to failure)

and recommend failure prevention methods

FA begins with non-destructive techniques, then proceeds to destructive techniques.

3) Introduction to Root Cause Analysis

- Section Summary

The Hierarchary:

Continously Improvemet

Essential to being compeditive and advancing objectives.

Problem Solving

An important method for continuous Improvement.

Root Cause Analysis

One type or problem solving approach that works to identify not only what and

how an undesired event occurred, but also why it happened so as to prevent

reoccurance.

Failure Analysis

A broad subcategory of Root Cause Anaylsis techniques that can be used when

failed or malfunctioning devices are available for examination.

FA has many sub categories and specialists realated to the type of technologies

and materials that failured.

4) RCA Approaches, Management & Reporting Methods

Root cause analysis is a generic term for

diligent structured problem solving.

Over the years various RCA techniques and management methods

have been developed

5 of the most popular RCA approaches are:

The” 5 Whys” Technique

The 8D (Eight Disciplines) Problem Solving Process

Shainin Red “X” Statistical Problem Solving

Six Sigma

Physics of Failure / Reliability Physics

4.1) The 5 Why Approach

1) WHY?

2) WHY?

3) WHY?

4) WHY?

5) WHY?

Mom, Why is the Sky Blue?

Why Can’t we see God?

Why is water wet?

Why . . .

4.1) The 5 Why Approach

The 5 Why’s is a simple problem-solving technique developed by

Toyota* to quickly get to the root of a problem.

The 5 Why strategy involves looking at any problem and asking:

“Why”? and “What caused this problem”?

The answer to the first “why” must prompt another “why”

and the answer to the second “why” must prompt another and so on.

The rule of thumb is that the “Why” question must be asked & resolved

at least 5 times in order to identify the true underlining root cause of the problem.

Toyota’s Philosophy:

A Rush to action that addresses only symptoms or a problem

only produces temporary relief.

Only after the “True” Root Cause has been identified can an

“EFFECTIVE STRATEGY TO PERMANENTLY RESOLVE”

the issue be developed.

* Ref. “The Toyota Way, by Jeffrey K. Liker, McGraw-Hill 2004

4.1) The 5 Why Approach

Example:

ISSUE

There is an oil spill on the floor

A machine is leaking oil

A gasket has failed

The gasket is made out of paper

which breaks down quickly

Low cost paper gaskets were purchased

instead of durable graphic or silicon gaskets

Purchasing bonuses are based on up front

cost savings not long term performance

OBVIOUS RESPONSE

This is a safety hazard, Clean it up

Fix the oil leak

Replace the gasket

Find a better gasket

Developed detailed specifications to

provide better direction to purchasing

Change purchasing incentive policy to

promote total value over short term savings

WHY

Toyota is known of not stopping at the technical issues. They continue until the root causes

of organization, cultural & people motivation issues are also understood & addressed.

4.1) The 5 Why Approach

The “Funnel Model”

Phase 1) Identify the Issue

Phase 2) Issue Investigation

Phase 3) Corrective Action

Vague Perception of Problem

Clarify Issues

Identify a Problem

or Concern

Issue

Obvious Cause

Cause

ROOT CAUSE

Develop Corrective Action

Evaluate / Verify Effectiveness

Implement & Standardize Across the Organization

WHY

4.1) The 5 Why Approach - Summation

Benefits

Easy to remember, Simple to apply,

Gets deeper into “Root Cause” than many other problem solving techniques,

so better in the long run.

Informal, flexible, open structure, little bureaucracy

Organizations/users adapts to their own needs.

Potential Issues / Concerns.

More time consuming investigate than quick fix approaches.

Sorting out issues with MORE THAN 1 CAUSE.

Mistakes in developing/answering a “Why” question can mislead the investigation.

Requires some Subject Matter Expertise

Hardest part of 5-Whys is asking the right “Why” questions.

Every organization does not have access to experts in every area.

Depends on some knowledge of cause & effect.

To ask the right questions,

Know how to follow them up in order to reach the right conclusions.

Novices can follow the wrong path.

Informal, flexible, open structure, little bureaucracy = Little guidance.

Repeatedly ask why can appear threatening to involved people.

Fear of an inquisition and assigning blame.

Self preservation instincts can lead to lack of cooperation or hiding information.

4.2) 8D (Eight Disciplines)

Team Problem Solving Process

8D is a problem-solving methodology that

emphasize team synergy.

Originated in 1974’s as part of MIL-STD-1520

“Corrective Action & Disposition System for Nonconforming Material”

Ford introduced and popularized the process within the Auto Industry in 1987.

First known as TOPS - “Team Oriented Problem Solving“.

Evolved into today’s widely used 8D process.

Philosophy - When a problem cannot be solved quickly by an individual,

a team approach is the most effect way to resolve the situation.

Team are more effective than the sum efforts of individuals working separately.

Essential to assign the right members to each a team and support them.

Team members need to have the inclination and skills needed for each problem

Team members need to be provided with the time and resourced need to solve the problem.

D4 – Determine & Verify

Root Causes

4.2) 8D Problem Solving Process (PSP)

Team

Appropriate

Problem

Identified

D1 - Select &

Empower a Team

D2 - Describe the

Problem

D3 - Implement &

Verify Interim

Containment

Actions

Identify Potential

Root Causes

Investigate and

Select Most Likely

Causes

the Potential

Cause a Real

Cause

Identify Potential

Corrective Action

D5 – Verify

Corrective Action

D6 – Implement

Permanent

Corrective Action

D7 – Prevent

Reoccurrences

D8 – Congratulate

the Team

Project Initialization Root Cause Investigation Implement

Corrective Acton

Yes

4.2) 8D PSP - Phase 1 Project Initialization

Starting Point - An Appropriate Problem is Identified.

8D Method does not define how problem awareness is developed.

Always use the right tool for the job:

Ensure problem warrant the resources of team PSP effort.

Avoid one size fits all tool and processes.

Avoid management dictates i.e. “all departments MUST deploy at least five 8D PSP per year”.

D1 - Use Team Approach

Establish a small group of people with the collective knowledge, time, authority and skills

to solve the problem, develop and implement corrective actions.

Provide each team with an executive champion to report to and clear roadblocks.

Each team requires a team leader to pace the process, lead meetings, coordinate team efforts.

Intermix skills: problem solvers, technical knowledge, manuf. process, test, analysis . . . etc.

Ensure team members have the inclination to work towards a common goal.

D2 - Describe the Problem

You can not fix a problem you don’t know what’s broke.

Clearly describe the problem in measurable, specific terms.

Clarify what, when, where and how much, impact to customers.

Info will be needed later to measure corrective action effectiveness.

D3 - Implement and Verify Short-Term Containment Actions

Stop or limit the bleeding as quickly as possible.

Define and implement screens, extra Q.C procedures, Rework . . . other appropriate actions.

To protect the customer & limit losses from the problem until a permanent C.A. is implemented.

Verify effectiveness with data and enhance if necessary.

4.2) 8D PSP - Phase 2 - Root Cause Investigation

4. Determine and Verify Root Causes -

Phase where team conducts the actual root cause Investigation.

Team applied experience and brain storm of preliminary information

to identify potential causes.

Team collects data, follows leads, performs analysis, authorizes test,

apply statistics . . . etc.

Specific procedures or tools not defined by the 8D process.

Team empowered to follow the facts, apply their expertise and

available resources to determine the best investigation approach.

Identification of “true” root cause(s) must be verified,

proven and documented by data not opinion)

to proceed to corrective action activities.

Concludes with team proposal for potential corrective action(s).

4.2) 8D PSP - Phase 3 - Corrective Action

5. Verify Corrective Actions -

Select the best case or optimal corrective action.

Perform test builds, process runs & evaluations to verify effectiveness & feasibility.

Confirm that the selected CA effectively resolves the problem without side effects.

Develop Corrective Action business case and obtain management approval.

6. Implement Permanent Corrective Actions -

Revise the product and/or process to implement the permanent fix

Establish monitoring to make sure it’s working.

If issues reoccurs implement additional controls or go back a few steps & try again.

7. Prevent Recurrence -

Improve practices & procedures to prevent recurrence of this & similar problems.

Modify specifications, update training, document lessons learned, review work flow.

8. Congratulate Your Team -

Recognize the collective efforts of your team.

Publicize accomplishments, share knowledge & learning across the organization

Going public with success spreads knowledge and learning.

Letters of thanks, certificates of recognition.

4.2) 8D Sample Reports

& Worksheets

Many 8D report templates exist.

Simple: “just the facts” & results

documentation reports

(Ref. Example right).

Complex: “document every step”

formats that include pages of

worksheets for preferred tools

(Ref. Following Example).

No Universal Format

Many format variations

possible.

Use what works for your

products, organization &

customers.

4.2) 8D PSP Variation

-The 5 Phase PSP

Simplified Version of the 8D.

Used to resolve & document less

complex / everyday issues.

That don’t require the resources

or expertise of a team approach.

Many Common Features:

1) Problem Description.

2) Immediate Actions.

2) Root Cause Conclusions.

RCA Investigation Plan Optional.

A Lesson Learned Opportunity?

4) Corrective Acton Plan (CAP)

Part / Process &

The System.

5) Verification/Validation of CAP.

No Universal Format

Many format variations possible.

Use what works for your

products, organization &

customers.

4.2) 8D/5 Phase Problem Solving Processes

-Summation

Benefits - Address a Number of 5 Why Concerns,

Early initial problem containment counter balance time need for thorough RCA

Drawing on team experience reduces potential for RCA errors.

Team format expands potential to tap available subject mater expertise.

Opportunity for novices to learn from more experienced personnel.

Provides a formal PSP structure without dictating methods and tools.

Team retain freedom to select tools and follows leads.

Team members feel empowered, respected and appreciate trust.

Easily converted into an 8D problem solving/ RCA report

Example 8D worksheet/ report template (on following pages) provides:

Sections for documenting outcome of all 8 steps.

RCA Worksheet for 5 Why and Fishbone Cause & Effect diagrams.

Status documentation

Potential Issues / Concerns.

8D structure provides susceptibility to excessive bureaucracy & micro-management.

Excessive status report updates detract from problem solving efforts.

Process management personnel represent non-value added overhead.

Management “throughput / efficiency” improvement efforts that degrade RCA effectiveness

(teams will avoid time consuming hard problems to avoid poor performance ratings

in systems that emphasize quantity over quality)

Management with lackluster team recognition / congratulations

4.3) Shainin Red “X”

- Diagnostic Journey

A Red X Statistical “Journeyman” or “Master” start the

process by organizing a team of problem stake holders.

The team creates a problem definition tree diagram

(similar to a fault tree minus the logic symbols).

Create a visual map of the issue or sequence of events

that relate to or resulted in a failure.

Included relevant issues & realistic contributing factors.

Use the diagram as a guide for evaluating the impact

of each factor.

Use progressive search questioning strategy a series of

(yes/no) questions concerning degree contribution to

reduce the field of suspects.

Cross off the factors that are minor contributors

to the outcome to eliminate them from serious

consideration.

The remaining factor in each category line are

considered to be the factors worthy of detailed

statistically evaluations.

Eliminate Source of

Cracking/Breakage of Vehicle

Exterior Rear View Mirror Glass.

Vehicle Relate Part Related

Feature Defect (27) Event

Loose

(3)

Inop. (4) Cracked

(11) NTF

(8)

Falls Off

(1)

Non Customer

Induced (4)

Customer Induced

(7)

Power Heated Mirrors

(DL3/DL8/DFP)(4) Other Mirrors

(0).

Region to Region

Same Glass Other

Strategies)

Crack in Other

Directions

Crack in East-West

Direction

Center of GlassTop of Glass . Bottom of

Glass

Driver’s Side Pass SideBoth

Thin Line Cracks Other Types of Crack

4.4) Six Sigma (6σ)

A methodology for “Improving Business Performance”.

Pioneered by Motorola Q.A. manager Bill Smith (mid 80’s) who proved that:

Manuf. lines with high in-process defects rates requiring Rework/Repairs (R/R) had higher

field failure rates & warranty costs than lines with low repair rates.

Low repair rate (build right on the first attempt) lines also had

improved customer satisfaction that resulted in better sales.

Root Causes:

Defect escapes from quality control systems.

Inadvertent, hidden damage during addition handling, rework & retest.

Lines with “better/tighter process capability” resulted in:

“Higher First Pass Quality” making them “More Efficient & Cheaper to Operate”,

even if the better equipment had higher up front costs, due to:

Less Effort & Costs for the “Hidden Factory” (Q.A, R/R & Root Cause).

Improved efficiency from higher throughput.

“Quality Pays” Even Better than Phil Crosby’s “Quality if Free” Philosophy.

Enabled QRD professional to communicate in the native language of executive

management: “Time and “$” Money”

Sigma σis the symbol for Statistical Standard

Deviation of the normal distribution (bell curve).

The “σ” measurement scale define how much

of process’s normal distribution is capable of

being contained within required tolerance limit

“ON THE FIRST PROCESS PASS”.

Out of spec “defects” are measured in terms of

Defects Per Million Opportunities (DPMO).

Processes that operate at a “6σ” quality capability

level produce < 3.4 DPMO “for each operation”.

DPMO is related to process operations not the

number of parts produced, Example:

A circuit board requires 100 component placement

operations so 1,000,000 placements ~ 10,000 boards.

The same board requires 500 solder joints so 1,000,000

soldering operations ~ 2,000 boards.

10,000 6σboards would require no more than

3.4 placement repairs & 17 solder repairs.

Wave soldering typ. run at 100-500 DPMO (4.78-5.19σ),

Reflow soldering is typ, 25-100 DPMO (5.55-4.78σ).

The goal is more capable processes that produce a

tighter variation spread within the spec limits

4.4) Six Sigma (6σ)

σ( Std Dev) Conversion Table

σIn Spec Yield DPMO (Outliers)

1 30.85% 691,462

2 69.14% 308,534

3 93.32% 66,807

4 99.38% 6,210

5 99.9767% 233

6 99.99966% 3.4

6σSpread

3σ

6σ

4σ

3σ

4.4) Six Sigma (6σ) Improvement Processes

DMAIC - Define, Measure, Analyze, Improve & Control

The 6σimprovement system for:

Existing” processes related problems

Sub-optimized process that fall below specification & yield expectations.

DMADV - Define, Measure, Analyze, Design & Verify

The 6σimprovement system for:

Developing new processes or products or

Resolving design related problems.

Also used in Design For Six Sigma (DFSS)

a methodology for new produce development.

Obvious similarities with the previously discussed 8D and 5 Phase PSP’s

Different definitions and terms.

Some differences in statistically tools.

4.5) Physics of Failure

- Definitions

Physics of Failure (PoF also known as Reliability Physics).

A Proactive, Science Based Engineering Philosophy.

Development & Applied Science of Product Assurance Technology base on:

A Formalized and Structured approach to Failure Analysis/Forensics Engineering

that focuses on total learning and not only fixing a current problem.

Material Science, Physics & Chemistry.

Variation Theory & Probabilistic Mechanics.

Up Front Understanding of Failure Mechanisms and Variation Effects.

Knowing how & why things fail is equally important to understand how & why things work.

Knowledge of how thing fail and the root causes of failures, enables engineers to identify

and design out potential failure mechanisms in new products and solve problems faster.

Provides scientific basis for evaluating usage life and hazard risks of new materials,

structures, and technologies, under actual operating conditions.

Applicable to the entire product life cycle

Design, Development, Validation, Manufacturing, Usage, Service.

4.5) PoF Grew Out of the Limitations of

Statistics Based Reliability Prediction

Fundamental Limitations

Statistical probability should be used only when we lack knowledge of the

situation and cannot obtain it at a reasonable cost.

"Statistics are applicable only when:

1. You are unavoidably ignorant about a given issue,

2. Some action is necessary and cannot be delayed."

Leonard Peikoff

In Book & Lectures on The Art of Thinking

In other words, if you're trying to determine a course of action:

- Your best bet is to acquire knowledge and

not to blindly use statistics to play the odds.

Focuses on 3 Separate Phases

with Separate Control & Improvement Strategies

4.5) A View of Quality, Reliability & Durability (QRD) Via

The Traditional Product Life Cycle Failure Rate “Bath Tub” Curve

Problem or Failure Rate

Time 0 1 Year 2 Years 3 Years 4 Years 5 Years

Average

Repurchase

Decision

The Bath Tub Curve

(Sum of 3 Independent Phenomena)

Durability = Wear Out

(End of Useful Life)

Reliability = Random or Chance Problems

(Constant Unavoidable)

But “True” Root Causes Can Be Disguised by

Statistical Assumptions to Make QRD Easy to Administer

This is an Inaccurate & Misleading Point of View

Quality = Infant Mortality

- Real failure rate curves are irregular, dynamic and full of valuable information,

not clean smooth curves to simplify the data plots.

4.5) A “PoF FAILURE MECHANISM” Based “REALISTIC” View

Reveals the True Interactive Relationships Between Q, R & D

Problem or Failure Rate

Time 0 1 Year 2 Years 3 Years 4 Years 5 Years

Manuf. Variation & Error

and Service Errors

That Cause Latent

Problems Throughout Life

“Cause & Effect”Root Causes

Can Be Disguised by Statistics

Once Problems Are Accurately Categorized

You Have a More Effective Point of View

Weak Designs That

Start to Wear Out

Prematurely

TRUE Random

Problems

Are Rare Once

Correlated to

“ACTS OF

GOD & WAR”

4.5) Root Cause Implications of the

Physics of Failure Point of view

The focus of “Traditional Reliability Methods” on “Random/Chance Failures”

conveys a perception that problems and failure are inevitable & unavoidable.

“Resistance is Futile”

The Physics of Failure approach emphasizes:

An ordered understandable, predictable universe of cause & effect relationships.

The role of root cause analysis problem solving for discovering, understanding

and mastering these cause and effect relationships.

Using RCA to build a “Compendium of Formalized, Institutionalized Knowledge”

for Future Problem Prevention as well as for solving today’s problems.



Failure Site :

The location of potential failures, typically the site of a designed in:

stress concentrator ,

design weakness or

(designed in)

material variation or defect.

(process related or Inherent)

Knowledge Used to Identify and Prioritized Potential Failure Sites and Risks

in New Designs During PoF Design Reviews.

4.5) Key PoF Terms and Definitions

GENERIC FAILURE CATEGORY TYP. FAILURE DETECTION

Errors - Incorrect Operations &

Variation Defects/Weaknesses.

Missing parts, incorrect assembly or process.

Process control errors (Torque, Heat treat).

Design errors

Missing functions,

Inadequate performance.

Inadequate strength.

Overstress.

Overheating.

Voltage/Current

Electro static discharge.

Immediate yield, buckling, crack.

Wearout/Changes,

via Damage Accumulation.

Friction wear.

Fatigue.

Corrosion.

Performance changes/parameter drift

4.5) 3 Generic PoF Failure Categories

and Detection Methods

Quality

Assurance

Immediate or

Latent defects

Performance

Capability

Assessments

Stress-Life

Durability

Assessments

4.5) Generic PoF Failure Categories

1) Overstress - When Loading Stress Exceed Material Strength

Variation of Design’s Material Strengths

- Related to Process Capabilities

Stress Variation of Usage &

Environments Loads &

Their Interactions

How well

do you

Understand

& Design

For

Strengths

& Stresses?

Typical

Deterministic

(Nominal) Analysis

FREQUENCY OF OCCURRENCE

STRESS/

STRENGTH

DESIGN MARGIN

SAFETY FACTOR

UNRELIABILITY = Probability

that Load Exceed Strength

4.5) Generic PoF Failure Categories

2) Errors and Variation Issues (They Are Everywhere)

People

Interface Equipment

Material Environment

Performance

Design & Process

Usage

Noise Factors

Errors Broadest Category

Errors in Design, Manufacturing, Usage &

Service.

Missing knowledge

Human factor Issues.

Variation

Fine line between excessive variation & out

right errors.

Both related to various quality issues.

Manufacturing equipment wear out & failure

could be related to maintenance errors.

Weak material could be raw material variation

or insufficient heat treat processing errors.

Equipment process capabilities limitation or

operator set up error.

4.5) Generic PoF Failure Categories

3) Wearout - Damage Accumulation In Materials

1. Loads

Elect. Chem.

Thermal, Mech...

Individual or

combined, from

environment &

usage act on

materials &

structure.

2. Stress

The distribution/

transmission of

loading forces

throughout

the device.

6. Time to 1st Failure:

(Damage Accumulation verses Yield Strength

A Function of: Σ [Stress Intensity, Material

Properties, & Stress Exposure Cycles/Duration].

7. Rate of Failure (Fall out)

A function of variation in; Usage, Device Strength

& Process Quality Control (i.e. latent defects).

3. Strain :

Instantaneous changes

(materials\structural) due

to loading, different loads

interact to contribute to a

single type of strain.

Knowledge of how/ which

“Key Loads” act & interact

is essential for “efficiently”

developing good products,

processes & evaluations.

4. Damage

Accumulation

(or Stress Aging):

Permanent change

degradation retained

after loads are removed.

From small incremental

damage, accumulated

during periods/cycles

of stress exposure.

5. Failure Site & Type:

Typically due to a designed in: stress concentrator ,

design weakness, material/process variation or defect.

4.5) Generic Failure Categories - Wearout (Damage Accumulation) con’t

3) Wear out Over Time and Intensity of Stress Exposure

How well

do you

Understand

& Design

For

Strengths

& Stresses?

INITIAL

UNRELIABILITY

FREQUENCY OF OCCURRENCE

STRESS/

STRENGTH

STRESS

EXPOSURE TIME

or USAGE CYC’S

Material Decay

Increases

UNRELIABILITY

OVER TIME

STRESS INDUCED

DAMAGE

ACCUMULATION

Design’s Strength

Decay/Spreads Over

Time / Usage

4.5) Generic Failure Categories

Overstress - Examples of Wear Out Failure Mechanism

Mechanical

Fatigue

Creep

Wear

Electrical

Electro-Migration Driven

Molecular Diffusion & Inter Diffusion

Thermal Degradation

When Over Stress Issue are Detected.

Verify supplier’s are meeting material strength specs & purity expectation.

Re-evaluate field loading / stress expectation used to design the part.

Sort out stresses,

Combined stress issues are often involved.

Re-evaluate effectiveness of product durability testing

Chemical / Contaminate

Moisture Penetration

Electro-Chemical-Migration Driven

Dendritic Growth.

Conductive Filament Format (CFF)

Corrosion

Radiation Damage

4.6) Physics of Failure Examples

- Circuit Board Related Vibration Durability Issues

Board in Resonance

Components. Shaken Off/Fatigued

by Board Motion.

By Flexing Attachment Features

Components In Resonance.

Components Shake/Fatigue themselves apart or off the

Board.

Especially Large, Tall Cantilever Devices

3 Med. Sized Alum CAPS

1 Small Long Leaded Snsr

1 Hall Effect Sensor.

1 Large Coil Assembly

PC Board

Lead Motion

-Flexed Down

- Normal

- Flexed up

Bending Lead Wires

Stressed

Solder

Joint

Displacement

Gull Wing I.C.

Time to Failure Determine by

Intensity/Frequency of Stress Verses Strength of Material

For a 10 million cycle life, Z < 0.0008995·B/(C·h·r (L1/2)).

Ref: Vibration Analysis for Electronic Equipment, by David S. Steinberg

Steinberg’s Criterion:

Log (Number of Cycles to Failure)

Log (Peak Strain)

Solder Fatigue Life

4.6) PoF Example

– E/E Module Vibration Analysis

Transformer

A Large Mass,

will drive a Large

Vibration Modal

Response

Connector Provides Primary PCB Support

Original CAE Guided Redesign

Adds Back Edge Support

Board Displacement (mils) 13.95 1.15

Natural Frequency (Hz) 89 489

Vib. Durability Calculation 25 Days > 50 Years

CAE Modal Simulation of Circuit Board Flexure

4.6) PoF Example Vibration Durability Calculations

-For Alternative PCBA Support & Mass Locations

ORIGINAL TRANSFORMER LOCATION

100

1000

10,000

100,000

10 M

100 M

1000M

Edge1

(Connector) Edge1 &

Corners Edge1 &

Middle Edge1,

Corners

& Middle,

Edge1 &

Edge2 All Edges

|| R101

+ R102

|| R825

+ R824

TRANSFORMER RELOCATED

Edge1

(Connector) Edge1 &

Corners Edge1 &

Middle Edge1,

Corners

& Middle,

Edge1 &

Edge2 All Edges

DAYS TO

FAILURE

@ 2 Hrs

Vib / Day

|| +

3650 Days

(10 Years)

Infrared Thermal Imaging Reveals Hot Spots From

Concentration of High Power Component

Surrounding Heat Sensitive Alum. Electrolytic Cap.

4.6) PoF Example

-Thermal Stress Balance/Distribution & Stress Avoidance

Another Design Uses an Array of Thermal Vias as a

Heat Spreader to Lower Peak Temperatures.

Alum. Caps Located Away From High Power Areas

Alum.

Caps

Alum. Caps

Away From

Heat

4.6) PoF Example - Moisture/Contaminate Failures

- Electro-Chemical Metal Migration Shorts - a.k.a. Dendritic Growth

Excessive Ionic Residue Contaminates on

Circuit Board can interact with atmospheric

humidity to form an electrolyte.

When a voltage differential is present

across a small distance copper ions can be

excited to migrate from the anode to the

cathode of the circuit ( + to -).

A copper trail will be deposited along the

way that will eventually support current

leakage short circuits.

4 factors are required:

1) Excessive Ionic Residues

2) Humidity (typ.>65% R.H. varies with Temp.

3) Exposed Copper.

4) Voltage difference bias over

a short distance

Ionic Chromatograph.

Identifies Electro-Chemical

Contaminates From Manuf. Processes

4.6) PoF Example - Moisture/Contaminate Failures

- Detrimental Contaminates

Chloride Residues

One of the more detrimental residues found on PCB

Typically related to flux residues.

Chlorides will initiate and propagate electrochemical failure mechanisms, such as

dendrite growth metal migration and electrolytic corrosion, when combined with

water vapor and an electrical potential.

Levels > 2 mg./sq. in. typically can not be tolerated.

Bromide residues

Generally related to bromide fire retardant in epoxy-glass laminates.

Can also come from solder masks, marking inks, or fluxes with bromide activators

Fire retardant, bromide is not typically degrading to long-term reliability of PCBs.

Bromide from a flux residue, can be very corrosive

Epoxy-glass laminate bromide levels typical range of 0 - 7 mg/sq. in.

Bromide levels >12 mg./sq. in. can be detrimental on organic PCB

Levels between 12-20 mg./sq. in. are borderline risks

Levels above 20 mg/sq. in. are a significant risk especially if from flux residues.

4.6) PoF Example - Moisture/Contaminate Failures

- Detrimental Contaminates

Sulfate Residue

Sulfates can come from many sources, contact with sulfur-bearing paper or

plastics, acid processes in fabrication, or from water used for rinsing & cleaning.

Minimal Risk: 0.0 – 1.0 mg./sq. in.

Marginal Risk: 1.0 – 3.0 mg./sq. in.

High Risk: > 3.0 mg./sq. in.

Sulfate levels above 3.0 mg./sq. in. are corrosive & detrimental to circuit reliability.

With sulfate levels above 3.0 mg/ sq in, look for a sulfate-bearing chemical used

in processing especially sodium/ammonium per-sulfate and sulfuric acid.

Nitrate Residue

Nitrate has approximately the same electronegative corrosivetivity as sulfate.

The mg./sq. in residue concentration risk levels for sulfate also apply to nitrate.

4.6) PoF Example - Moisture/Contaminate Failures

- Detrimental Contaminates

Wear Organic Acids (WOA)

WOAs like adipic or succinic acid, are activators in many solder fuxes

Residue levels vary greatly with the flux delivery system (foam, spray, paste) and

the heating profile the determines the rate of consumption during soldering.

Low solids solder paste: 0-20 mg./sq.in.

Spray-applied, low-solids flux: 20-120 mg./sq.in.

Foam-applied flux process: 120-150 mg./sq.in.

Water soluble flux w/good cleaning: 0-15 mg./sq.in.

Water-soluble fluxes generally have a much lower WOA content than low-solids

(no-clean) fluxes.

WOA levels are under 150 mg./sq. in. are generally not a risk.

Excessive WOA amounts (>150 mg/in2) present a significant PCB reliability risk.

Un-reacted WOA flux residues will readily absorb atmospheric moisture then

support corrosion and the formation for current leakage dendritic growth failures.

4.6) PoF Example

- Capacitor Flex Cracking Examples

Capacitor

Ends Bend Down

Tensile Stress (Crack Site) on Top

Ends Bend Up

Tensile Stress (Crack Site) on Bottom

Cracked

Chip Cap,

Capacitor are fundamental, passive electric devices for

energy/electron/charge storage.

A cap is formed by two parallel conducting plates

(electrodes) separated by a dielectric material.

- Dielectrics are insulators, poor conductor of electricity

that support electrostatic fields. Rather than passing

an electric current, dielectrics absorb electronics into

an electro-static field.

For solid dielectrics such as Barium Titanate (BaTiO3) a

hard, brittle ceramic, many small plates/dielectric

sections are stacked in parallel to create a large

capacitance in a very small package.

The brittle fragile nature of the thin dielectric ceramics

can result in fracture cracks in the capacitors if their

circuit boards experience occasional bending or flexing.

4.6) PoF Example

- Thermal Stress & Thermal-Mech. Durability

Durability Simulations Identifies Most Likely

Parts to Fail Due To Thermo-Mechanical Fatigue

Identified (Large Body 1020-S.M. Resistors)

Thermal Analysis

Identifies Internal

Thermal Stress &

Overstress “Hot Spots”

From Power Dissipation

& Environment

Conditions.

1020 Resistor Fatigue Confirmed

In Accelerated Life Test

Infrared Thermal Imaging Of Thermal

Stress & Overstress “Hot Spots”

Thermal-Mechanic Durability Modeling to Identify Potential

Intermittent Circuits Due to Themo-Mechanical Fatigue

5) Manufacturing Issues

Highly Reliable Products Need To Be

Built Right As Well As Designed Right.

A Robust Well Balanced Design Can Be Rendered

Un-Reliable by Fabrication and Assembly Errors or

Excessive Variation Issues.

A Consistent and Capable Manufacturing Process and

Supply Chain is also Required

5) Manufacturing Issues

The 5 Most Common E/E Device Manufacturing Issues

Rework & Repair

Latent Rework & Handling

Damage (% Varies)

RE-HEAT,

REWORK &

REPAIRS

In Process Board Flexure

Cracked & Missing Components.

(Related to up to 15% Of E/E Assembly Issues).

Ionic Contaminate

(Circuit Board Cleanliness to Prevent

Humidity Related Short Circuit Growths)

(Related to up to 20% Of E/E Assembly

Issues).

ASSEMBLY & SOLDERING PROCESS

(Related to up to 60% of E/E Assembly Issues)

6 Sigma

Electro Static Discharge (ESD)

(Component Damage)

(% Varies Often Related To Spills)

Most Root Cause techniques are only call upon after a failure has happen to determine what when wrong.

But the many of same methods can also be used to determine if new products are being built right

6) Identifying What the Problem Is

- Root Causes Failure Analysis Techniques

Return parts Root Cause Failure Analysis always starts with

Non-Destructive Evaluation (NDE).

Designed to obtain maximum information with minimal risk of damaging or

destroying physical evidence

Non Destructive Evaluation Methods

Visual Inspection

Electrical Characterization

Optical Microscopy

Scanning Electron Microscopy

Acoustic Microscopy

Xray Microscopy

Infrared Thermal Imaging

SQUID Microscopy

Spectral Material Analysis

(Elemental Composition)

Ion Chromatography-Chemical Analysis

Destructive Evaluation Methods

Decapsulation

Microsectioning

Metallographic Metallurgical Analysis

Focused Ion Beam Milling

Electrical Transient Probe Testing

Material Property Characterization

Thermo Mechanical Analysis (TMA)

Differential Scanning Calorimetry (DSC)

Polymer Thermal-Mechanical Properties

6.1 Visually Aided Inspection

- Microscopy Optical & SEM

Enables the visualization, inspection and evaluation of tiny

objects and details.

Light based optical Microscopes provides magnifications

up to 1500x, resolution down the 0.2 micrometer.

Electron beam based Scanning Electron Microscopes

provides magnifications up to 2,000,000x.

Modern professional grade microscopes are equipped with

digital imaging capture for documentation and comparison

purposes.

6.1) Microscopic Failure Analysis of Solder Separation in BGAs

- Root Cause: Excessive Underfill Thermal Expansion

6.1) Microscopic Failure Analysis of Solder Joint Fracture

- Root Cause: Failure Due to Gold Embitterment

Cross section of failed solder joint revealed excessive Gold-Tin (AuSn4) intermetallics.

SEM Energy Dispersive X-ray Spectroscopy (EDS) found solder’s gold content >8%.

Embrittlement will occur if gold content exceeds 3.5% by weight.

Excessive component gold plating allowed large amount gold to diffuse into the solder .

Controlling Factors: Excessive Gold, Soldering Temperature and Time Above Liquidus

1200x SEM image reveals needle-like

structures of AuSn4intermetallics in

the solder joint

Cross section of component with

thickness of gold plating layer.

6.2) Thermal Imaging Microscope

Thermography is the use of an infrared imaging

and measurement camera to "see" and "measure

" thermal energy emitted from an object.

Provides precise non-contact temperature

measurement capabilities.

Spectral range can be broken into one of four

ranges, near IR: 0.75-3 microns, middle IR: 3-6

microns, far IR: 6-15 microns and extreme IR: 15-

30 microns.

Important parameters include measurement

temperature range, spectral range, accuracy,

resolution and steady state vs. real-time

Resolution, PCBA: 15 microns

Resolution, on-die: 1 micron

Use points

Find Electrical shorts

Power Components

Identify Temperatures,

Find Hot Spots

Trace Heat Flow Paths

6.2) Infrared Thermal Imaging

D11

Q16

Thermal Anomalies Detected

- Q16 producing heat when is it suppose to

be in an off state - Sneak circuit detected.

- D11 detected a hot spot that exceeded

thermal bogies.

Resulting in over heating near by Alum Caps

Alum Caps

6.3) X-ray Microscope

Enables internal inspection through the use

of X-ray energy

Latest innovations

Digital Detector

Laminography (‘virtual’ cross-sectioning)

3D reconstruction

Nanofocus resolution

Oblique viewing

6.4 Acoustic Microscopy

Non destructive method for inspecting internal structures.

By mapping the echo pattern of high frequency (>20 kHz) sound waves.

Sonic energy excites loose or moveable structures.

Requires immersion in water (acoustic signals reflected by air)

Enable non-destructive detection/location of:

structures, cracks, voids and delamination

H2O

Transducer

Receive

6.5) SQUID Microscopy

Superconducting QUantum Interference Device

Current flow in devices produce a magnetic field

SQUID uses a highly sensitive magnetic detector

(superconductor) to resolve these fields

Magnetic field image is converted to a current

density image, allowing for fault location

Resolution down to 300 nm

Critical technology for detecting the current path

of electrical shorts through a package or material.

6.6) Micro Cross Sections - a Destructive Analysis technique for the

internal evaluations of component's good for detecting manuf. defects

-Metallographic Analysis involves X-Sections of metals

(i.e. Leads & Solders) for material quality evaluations.

Thru Hole Pins

Text Book Perfect

This Webinar is a based on a 2 day Short Course:

“Understanding Failure & Root-Cause Analysis in Electronics”

1) Introduction and Objectives

The Need for Root Cause Analysis

Difference Between Problem Solving, Failure Analysis

& RCA Of Field Failures

2) Root Cause Approaches, Management &

Reporting Methods

The” 5 Whys” Technique

The Eight Disciplines (8D) Technique

Shainin Red “X” Statistical Problem Solving

Six Sigma

Physics of Failure/ Reliability Physics

Break

3) Generic Failure Categories

Design Quality & Errors

Manufacturing Quality & Errors

Environmental & Usage Considerations - Their Role

in Over Stress & Accelerated Wear Out Failures

Environment & Self Heat Temperature Issues

Vibration, Shock & Drop

Humidity

Contaminates

4) Finding Failure Modes – Where Problems Are &

How They Manifest Themselves.

The Need for Data

Collecting & Analyzing Data for Problem Solving

Trending analysis results (plotting a timeline

Pareto Analysis

Other Data Sources

Test Reports

Warranty Data

Fleet Maintenance Logs/Reports

Customer Surveys

Investigation Interviews

Lunch

This Webinar is a based on a 2 day Short Course:

“Understanding Failure & Root-Cause Analysis in Electronics”

5) Fault/Failure Investigation - Identifying

What the Problems Is

- Part I) Developing a Hypothesis

Customer & Service Technician Feedback &

Interviews,

Reference Product & Technology History/Lessons

Learned

Identifying Contributing Events

Ishikawa (fishbone) diagrams

Fault Tree Analysis

Dealing with Multiple Problems – Event/Issue

Charting

6) Identifying What the Problem Is

- Part II) Return Parts Analysis

Managing a Return Part Program

Initial Issue Confirmation Functional Checks

Electrical Fault Isolation

Break

7) Identifying What the Problem Is

- Part III) Root Causes Failure Analysis

Physical Component Failure Analysis Laboratory

Methods

Cross-Sectioning / Metallographic Analysis

IC Decapsulation

Optical Microscopy

Electron Microscopy

Ion Chromatography

Surface Analysis (FTIR, EDS, XRF, etc.)

Material Analysis (DSC, TMA, TGA, etc.)

Mechanical Analysis Techniques

(Micro-tester, Bend Testing, Pull Testing, etc.)

End of Day One

This Webinar is a based on a 2 day Short Course:

“Understanding Failure & Root-Cause Analysis in Electronics”

8) Typical EE Failure Modes, Mechanism &

Signatures

Printed Circuit Board Substrate Issues

Manufacturing Defects

Plated Through-Hole Via Issues

Conductive Anodic Filaments

Electrochemical Migration (Dendritic Growth)

Integrated Circuit Packaging & Die Issues

Wire Bond Failures

IC Pop Corning

Integrated Circuit Die Issues

ESD/EOS

Fluid Penetration Issues (new)

Thermal Issues (new)

Break

Capacitors (Ceramic, Aluminum, Tantalum)

Passive Components

Electro-Mechanical Devices.

Terminals and Contacts

Wire Failures

Relay

Speakers & Audio Alarms PCB Assembly

Solder Quality Issues

Lunch

9)Using CAE Simulation in RCA

Vibration & Shock

Thermal Simulations

10) Developing/Implementing a Permanent Corrective

Action Plan

Developing the Corrective Action Plan

Stakeholder Teamwork & Buy In.

Fixing the Problem Rather Than Assigning Blame

Fixing the Design, the Supply Chain or Assembly Process

Building a Business Case/Getting Approval for the Plan

Internal Failure Rev. Board/Management Rpts & Approval

Customer Reports and Approval

Regulatory Agency Review & Approval

Break

Validating the Fix

Implementation Verification

Learning From Failure - Corrective Action to Prevention

Documenting the Issues

Document and Reusing Lessons Learned

Implementing the Fix

Engineering and Validation Issues

Assembly Processes, Maunf. & Quality Issues

Suppliers and Supplier Quality Issues

Wrap-Up & Adjourn

Want to Know More,

Contact Your Nearest DfR Solutions Location

Midwest Office

Jim McLeish

248-726-7600

jmcleish@dfrsolutions.com

Bay Area Office

John McNulty

415-806-7704

jmcnulty@dfrsolutions.com

Sales Manager, Southwest

Clayton Bonn

cbonn@dfrsolutions.com

Corporate Headquarters

College Park, MD

301-474-0607

askdfr@dfrsolutions.com

Page 1

Clifford H. Lange

Proprietary Document

Root Cause Analysis

Mechanical Components and Systems

by Clifford Lange, PhD, PE, Ops A La Carte

Clifford H. Lange

Proprietary Document

Page 2

Root Cause Analysis – Mechanical Components

Polling Questions

Are you familiar with creep related problems or have direct

experience with solving a creep issue?

Don’t know what creep is

Some familiarity with creep

Direct experience with creep behavior

Do you understand the application of structural reliability

methods (e.g. FORM/SORM) for the understanding of failure

mechanisms

Don’t know what structural reliability methods are

Some familiarity with structural reliability methods

Direct experience with structural reliability methods

Clifford H. Lange

Proprietary Document

Page 3

Design for Reliability – Mechanical Components

Conform to accepted industry design standards (ASTM, SAE,

ANSI, etc.)

Avoid the need to use high tolerances (e.g. < 0.010”) and be

cognizant of tolerance stack up issues

Ensure compliance with all recommended rating guidelines

Anticipate unusual environmental effects

Incorporate contract manufacturers early in the design process

(they are the experts)

Perform reliability assessment on primary wearout

mechanisms

Clifford H. Lange

Proprietary Document

Page 4

Critical elements of mechanical systems

Transmitting elements

Shafts, belt drives & flexible couplings

Springs & gears

Actuators, accumulators & reservoirs

Brakes & clutches

Motors, pumps & valves

Constraining, confining, & containing elements

Seals & gaskets

Bearings & Shaft sealing devices

Fixing elements

Bolted connections or threaded fasteners

Weldments

Elements supporting machinery functions

Lubrication systems

Clifford H. Lange

Proprietary Document

Page 5

Typical failure mechanisms of mechanical systems

Stress rupture or fracture

Insufficient design

Changes in load history or component application

Fatigue

Poor material characterization or load history

Creep

Wear and/or fretting

Environmental effects

Corrosion

IGSCC

Hydrogen embrittlement

Clifford H. Lange

Proprietary Document

Page 6

Reliability prediction for mechanical systems

Bloch, H.P. and Geitner, F.K.; “An Introduction to

Machinery Reliability Assessment;” Van Nostrand

Reinhold, 1990.

“Handbook of Reliability Prediction: Procedures for

Mechanical Equipment;” Naval Surface Warfare

Center – Carderock Division; CARDEROCKDIV,

NSWC-94/L07, March 1994.

Clifford H. Lange

Proprietary Document

Page 7

Example: Creep Failure

High temperature aluminum heater weldments

Pre-stressed concrete (water) pipe failures

Power plant steam pipe creep rupture

Steam pipe ruptures lead to in depth inspections at all aging

facilities

Main steam piping at TVA Gallatin Units 3 & 4 showed

excessive deformation (~ 10% radial strain – wall thinning)

Average diametral strain is 5.3% (swelling)

Initial “thin-wall” creep calculations indicated evidence of

bending moments but results were inconsistent with data

Thick wall “finite element” calculations improved predictions

Results indicated that the ASTM creep rate law predicts

approximately 2x service heater data

Clifford H. Lange

Proprietary Document

Page 8

8.6 Example: Creep Failure of Steam Piping

Clifford H. Lange

Proprietary Document

Page 9

8.6 Example: Creep Failure of Steam Piping

Comparison of Wall Thinning Comparison of Diametral Swelling

Results reflect ASTM Creep Rate Law

Thick Wall creep results are more

consistent with measured diametral

swelling

Thick Wall creep results are more

consistent with measured wall thinning

Clifford H. Lange

Proprietary Document

Page 10

8.6 Example: Structural Reliability as a RCA Tool

Wind Turbine design provides a good example of an

ongoing RCA program

Traditional fatigue analysis often focus on uncertainty with the

material properties and/or the load (e.g. stress) spectrum

New technology (e.g. Structural Reliability Methods)

employed to improve the RCA

In many cases uncertainty in the underlying load environment, the

stress response and the computational techniques employed can be

significant contributors to fatigue failures

Problems involving many different sources of uncertainty are

effectively addressed using Structural Reliability Techniques

Clifford H. Lange

Proprietary Document

Page 11

8.7 Example: Fatigue – Traditional Analysis

Wind turbine blade application

Typical S-N data for aluminum used for design

Stress spectrum assumed to be determined

experimentally – Monte Carlo simulation used

to generate sample stress distribution

Fatigue analysis considers both best fit and

95% CI on S-N properties as well as the

measured stress histogram and a bounding

load spectrum

Results compared across all assumed input

variables

Clifford H. Lange

Proprietary Document

Page 12

8.7 Example: Fatigue – Material Behavior

Fatigue data is for 6063

Extruded Aluminum

Both a least squares best

fit and a 95% confidence

level used in fatigue

analysis

Miner’s Rule used to sum

fatigue contributions

over different stress

amplitudes

1.70

1.80

1.90

2.00

2.10

2.20

2.30

2.40

2.50

2.60

2.70

2.80

1.0E+04 1.0E+05 1.0E+06 1.0E+07 1.0E+08 1.0E+09

Cycles to Fail

Stress (MPa)

Least Sq. Fit, C=5.00E21

95% Lower CI, C=9.52E20

Teledyne Engr.

Runout Specimen

Southern Univ

100 N=Cσ-b

b= 7.3

250

500

1=+⋅⋅⋅⋅++=Δ

Clifford H. Lange

Proprietary Document

Page 13

8.7 Example: Fatigue – Applied Stresses

Applied stresses for wind

turbine blade vary with

wind speed

A typical wind speed

distribution representative

of mid-west USA is

assumed

Distribution is Weibull with

α= 2.0 & μ= 6.3 m/s

5 different stress amplitude

distributions are assumed

for 5 corresponding wind

speed bins between 0 and

25 m/s.

[]

exXP

⎥

⎦

⎤

⎢

⎣

⎡

−

−=≤ 1

()

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0 5 10 15 20 25

Wind Speed (m/s)

fX(x)

Bin 1 Bin 2 Bin 3 Bin 4 Bin 5

Clifford H. Lange

Proprietary Document

Page 14

8.7 Example: Fatigue – Applied Stresses

Distribution of stress

amplitudes stresses in

each wind speed bin

also assumed Weibull

Assume αs= 2 with

shape factor βslinearly

dependent on wind

speed, X

Contribution potential

for high stress

amplitudes is evident

[]

exsXSP

⎥

⎦

⎤

⎢

⎣

⎡

−

−=≤

1||

⋅

2.1

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0 102030405060

Stress (Mpa)

fS(s)

Wind Speed Bin 0 - 5 m/s

Wind Speed Bin 5 - 10 m/s

Wind Speed Bin 10 - 15 m/s

Wind Speed Bin 15 - 20 m/s

Wind Speed Bin 20 - 25 m/s

Clifford H. Lange

Proprietary Document

Page 15

8.7 Example: Fatigue – load Spectrum

Monte Carlo simulation

used to produce 10K

stress amplitudes

Assumed design load

spectrum used to

model anticipated long

term loading conditions

Both histogram and

load spectrum used in

analyses

0.00001

0.0001

0.001

0.01

0.1

1 7 13 19 25 31 37 43 49 55 61 67 73 79 85

Stress (MPa)

Relative Frequency

Assumed Design

Load Spectrum

Clifford H. Lange

Proprietary Document

Page 16

8.7 Example: Fatigue – Risk Level?

All 4 combinations of C and Loading used to evaluate

relative influence of each parameter & uncertainty level

Both fatigue lifetime and damage results presented

Results show satisfactory design against fatigue failure

S-N

Material Loading Lifetime:

years Damage

Cave Histogram Data 1232 .0162

Cave Design Spectrum 426 .0470

C.95 Histogram Data 216 .0925

C.95 Design Spectrum 81 .2465

Clifford H. Lange

Proprietary Document

Page 17

8.7 Example: Fatigue – Structural Reliability

Used to evaluate designs probabilistically considering both

the mean and standard deviation of design inputs

Results are probabilities of failure and the relative

importance of each input (random variable)

For fatigue – rather than ask;

“What is the actual fatigue life of the component?”

the more appropriate question;

“With what confidence will the component meet it’s

target lifetime?”

can now be answered.

For RCA we can identify the leading contributors to failure

Clifford H. Lange

Proprietary Document

Page 18

8.7 Example: Fatigue – Structural Reliability

Intuitively the risk or probability of failure can be inferred from the

overlap of the region of the load and resistance random variables

Both the relative values of the mean and variance of each random

variable affect the failure probability

σx2

σx1

x2 (Load) μx1 (Resistance)

Clifford H. Lange

Proprietary Document

Page 19

U2 =

U1 =

g(U) = 0

−

Design Point

U2 =

U1 =

g(U) = 0

−

Design Point

8.7 Example: Fatigue – Structural Reliability

Limit state equation,

G(X), defines the fail

and non-fail conditions

Failure probability

determined by the μ

and σ2of X1and X2

Calculations performed

in standard “U-space”

where the design point

determines both the pf

and the relative

importance of X1& X2

)(

−

Φ=

σσ

μμ

−

(

)

21 XXXG

−

Clifford H. Lange

Proprietary Document

Page 20

8.7 Example: Fatigue – Structural Reliability

In the general formulation

the limit state equation is

not linear and the random

variables are not Normal

Linear (FORM) and

parabolic (SORM)

approximations are used

at the design point to

calculate failure

probabilities and

importance factors

g(U) > 0

g(U) < 0

SORM

FORM

g(U) > 0

g(U) < 0

SORM

FORM

Clifford H. Lange

Proprietary Document

Page 21

8.7 Example: Fatigue – Structural Reliability

Limit State Equation defines failure conditions

Time to failure determined using Miners rule with an average

damage per cycle

Average damage rate determined considers all possible stress

amplitudes and their incremental damage

()()

()

∫∫

∞

xs f

XXS dsdx

xfxsf

Tf0

(

)

tf TTXG

−

Clifford H. Lange

Proprietary Document

Page 22

8.7 Example: Fatigue – Structural Reliability

Both the underlying environmental variable, X, and the stress

amplitude, S, given the load environment, are Weibull distributions

With the shape factor, bX, of the environment determined from the

average, X, and the average of the stress response dependent upon

the environment

[]

exXP

⎥

⎦

⎤

⎢

⎣

⎡

−

−=≤ 1

[]

exsXSP

⎥

⎦

⎤

⎢

⎣

⎡

−

−=≤

1||

()

Clifford H. Lange

Proprietary Document

Page 23

8.7 Example: Fatigue – Structural Reliability

The RMS of the stress

process is a function of

the underlying

environment variable, X

Random vibration theory

is used to define the

shape factor, βs, as a

function of the RMS

stress and shape factor,

The RMS exponent, p,

used to identify

increasing/decreasing

stress processes

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

0.00.51.01.52.0

Load (e.g. WInd Speed)

RMS Stress (MPa)

0.00

0.62

1.25

1.87

2.50

3.12

3.74

4.37

Average Stress (MPA)

σref = 1.75

Xref = 1.0

p=1

p>1

p<1

⎟

⎠

⎞

⎜

⎝

⎛

=ref

ref

σσ

)(

2/1

])!/2/(2)[( ss x

ασβ

Clifford H. Lange

Proprietary Document

Page 24

8.7 Example: Fatigue – Generalized Formulation

Resulting expression for fatigue life a function of 12

random variables

Stress parameters K and σref are raised to the power, b, as

a result of the S-N relationship

Environmental parameters, X, are raised to the composite

power, bp, reflecting the combined nonlinear effect of the

RMS stress on the environmental variable, X

⎥

⎦

⎤

⎢

⎣

⎡

⎟

⎠

⎞

⎜

⎝

⎛

⎟

⎠

⎞

⎜

⎝

⎛

⎟

⎠

⎞

⎜

⎝

⎛

⎟

⎠

⎞

⎜

⎝

⎛

−

=!!

)!/1(

)/1()!/2(

0xs

xref

ums

ref

fbpb

SKS

ααα

Clifford H. Lange

Proprietary Document

Page 25

8.7 Example: Fatigue – Traditional Approach

CYCLES computer

program used to perform

calculations

Input values reproduce

those used in the traditional

fatigue analysis

Results confirm previous

results that fatigue design is

not likely to fail

Most significant input is the

S-N coefficient

Var Definition Dist Type Mean COV

X Mean Wind Speed Constant 6.3 -

αx Wind Shape Factor Constant 2.0 -

xref Ref Wind Speed Constant 1.0 -

σref Reference Stress Constant 1.75 -

p RMS exponent Constant 1.0 -

K Stress Conc Factor Constant 1.0 -

αs Stress Shape Factor Normal 2.0 .15

C S-N Coefficient Weibull 5E21 .613

b S-N Exponent Constant 7.3 -

fo Cycle Rate Constant 1.2 -

Δ Miner’s Damage Constant 1.0 -

Mean Lifetime: 467 years

Failure Probability FORM .61 %

SORM .94 %

Importance

Factors: Stress Shape Factor: 24.9 %

S-N Coefficient, C: 75.1 %

Clifford H. Lange

Proprietary Document

Page 26

8.7 Example: Fatigue – Generalized Approach

There exists uncertainty

in design inputs other

than the S-N law and

loading spectrum in

fatigue design

X, ax, sref, p and K are all

considered to be

uncertain in the wind

turbine example

Uncertainty in Miners

rule and the fluctuating

cycle rate are also

considered

Var Definition Dist Type Mean COV

X Mean Wind Speed Normal 6.3 .075

αx Wind Shape Factor Normal 2.0 .15

xref Ref Wind Speed Constant 1.0 -

σref Reference Stress Normal 1.75 .075

p RMS exponent Normal 1.0 .05

K Stress Conc Factor Normal 1.0 .1

αs Stress Shape Factor Normal 2.0 .15

C S-N Coefficient Weibull 5E21 .613

b S-N Exponent Constant 7.3 -

fo Cycle Rate Normal 1.2 .2

Δ Miner’s Damage Normal 1.0 .15

Clifford H. Lange

Proprietary Document

Page 27

8.7 Example: Fatigue – Results

Considering uncertainty

contributions from all

potential sources

changes the conclusions

from the original analysis

Failure probabilities have

increased to

unacceptable levels (5-

10%) while the mean

lifetime remains

unchanged

Most significant inputs

are mean wind speed and

the RMS exponent, p

Mean Lifetime: 467 years

Failure

Probability

FORM 5.67 %

SORM 7.38 %

Importance

Factors:

Mean Wind Speed, X: 6.7 %

Wind Shape Factor, αx: 25.2 %

Reference Stress, σref: 6.2 %

RMS exponent, p: 24.0 %

Stress Conc Factor, K: 10.6 %

Stress Shape Factor, αs: 9.3 %

S-N Coefficient, C 16.6 %

Cycle Rate, fo: 0.8 %

Miner’s Damage, Δ 0.5 %

Clifford H. Lange

Proprietary Document

Page 28

8.7 Example: Fatigue – Structural Reliability

Structural Reliability methods provide risk levels

(e.g. pf) as well as the relative importance of the

design inputs (e.g. random variables)

All 3 aspects of the fatigue problem; the loading

environment, structural response and the local

failure criterion may include uncertainty and can be

included in the fatigue evaluation

The methodology can employed through alternative

limit state equations or extended to other fatigue

problems (e.g. crack growth).

The most critical design inputs are identified

RCA Case Study

The PEPCON Incident

A Process Plant Accident &

Guidelines for General Investigations

T. Kim Parnell, PhD,PE

Root Cause Analysis Webinar

July 23, 2008

Why Review a Plant Accident?

• Interesting and well-studied event

• Provides general guidelines for RCA team

organization

• Insights for investigation and

documentation

• Contrast investigation of “unique” event

like this with RCA of high-volume products

The PEPCON Incident

• Fire and massive explosions at the PEPCON

plant in Henderson, NV on May 4, 1988.

• PEPCON produced Ammonium Perchlorate

(AP) – an oxidizer

• Combination of events:

– Human error – cigarette likely started initial fire

– Large quantity of AP on site due to Challenger

disaster

– 16” natural gas line running under the plant (with

leaking stitch welds)

PEPCON Explosions

• Two large explosions equivalent to 200

Tons and 500 Tons of TNT (3.0 and 3.5 on

the Richter scale)

• Over $70M property damage; windows

broken up to 30 miles away

• 16” Natural Gas Pipeline

– Ruptured 40 foot section

– Crushed more than 260 feet

– Long-term leakage prior to blast from poor

stitch welds

PEPCON Incident Investigation

• Organization of Teams

• Site documentation and evidence

collection; develop timeline

• Metallurgical analysis; Fracture mechanics

• Fire cause & origin

• Gas migration through soil

• Blast effects & damage

• Conditions for AP deflagration/detonation

Fire & Brimstone

• Rapid spread of fire;

catastrophic explosion

• Most of event captured on

video

Massive Explosion & Shockwave

• Stills from video shot from Black Mountain – over 10 miles away

Aerial View - Before & After

Before After

Near Ground Zero…

• Rail cars overturned

• Autos overturned

At the Plant

Plant Buildings

Production Equipment

Ruptured Gas Pipe – Initial View

Pipe After Some Digging

Pipe After Complete Excavation

Ruptured & Crushed Pipe Sections

Pipe Sections

16” Natural Gas Pipeline

• Ruptured 40 foot section

• Crushed more than 260 feet

• Long-term leakage prior to blast from poor stitch

welds

• Big Question: Did the pipe rupture occur before

or after the explosions??

Pipeline Section Identification

Pipe/Soil Finite Element Model

Reference:

Parnell, T.K. and Caligiuri, R.D., “Analysis of the Dynamic Response of a Buried Pipeline

due to a Surface Explosion,” Computational Aspects of Impact and Penetration, L. E.

Schwer and R. F. Kulak, eds., Elme Press International, 1991.

Pipe Crushing Due to Blast

Response Sequence #1

Pipe Crushing Due to Blast

Response Sequence #2

Pipe Crushing Due to Blast

Response Sequence #3

Pipe Crushing Due to Blast

Response Sequence #4

Pipe Crushing Due to Blast

Response Sequence #5

Pipe Crushing Due to Blast

Response Comparison

Pressurized Depressurized

Summary

• Document in detail

– Inspection

– Measurements

• Get the right expertise on the team; update

as needed

• Develop the scenario

• Test the hypotheses

PEPCON Explosion - References

• Video links

– History Channel, 8:00 minutes

http://video.aol.com/video-detail/pepcon-explosion-may-

1988/1249549102

– Exponent – 2:00 minutes

http://www.youtube.com/watch?v=HJVOUgCm5Jk

–Z-Axis

http://podcasts.zaxis.com/pac/pepcon-explosion

– Summary article

http://www.interfire.org/res_file/pdf/Tr-021.pdf

http://www.reviewjournal.com/news/pepcon/

Data Driven RCA

Al Alaverdi

SigmaQuest

Solutions for Data Driven Quality

Management & RCA

Focus on High Tech, Telecom,

Consumer Electronics, Medical

Devices

Good Data = Shortest path to RCA

RCA – A 360°Perspective

Engineering

Manufacturing

Customer Complaints &

Returns

Component

Suppliers

Install Base

Feedback

Eliminate Data

Fragmentation

“Single View of Truth”

Data Acquisition Challenges

Political

Engineering, Ops, Service, Quality

Component Suppliers, CMs, Repair Centers

Data Quality

Are you collecting the right data ?

Accuracy, Granularity, Latency

Consistency (Part #, Serial #, Version, Revision)

IT

Data Storage, Analytics , Large volumes of data

Typical Scenario - Data,

Data Everywhere

Customers Oracle

RCA

3PL-1

(Module)

3PL-2

(Module)

3PL-1

(System)

Oracle

Quality

Corporate

Reliability

Reports

IFS /

Stars /

Tars

XML

(7C6)

Pass/Fail Data

Desktop

Database

Dept.

Database

Test Data

Part B

Test

database

Detailed

Test Data

Part A

Call Home

Screening / Failure

Analysis Centers

40% of customers

File Incident Ticket

RMA

Receipt

0.5%

99.5%

SAP

Repair-1

(Component)

Repair-2

(Component)

Repair Centers /

sub-tiers

Repair/

FA data

Home Grown Repository (OEM)

Supplier Repository

ERP / CRM software (OEM)

Legend

Other

Reports

Test & Repair

SQE FA

Reports

Internal

RMA

(System)

Engineering

Feedback

Supplier

Root Cause

FA data

Repair/

FA data

Partial Data

Customers Oracle

RCA

3PL-1

(Module)

3PL-2

(Module)

3PL-1

(System)

Oracle

Quality

Corporate

Reliability

Reports

IFS /

Stars /

Tars

XML

(7C6)

Pass/Fail Data

Desktop

Database

Dept.

Database

Test Data

Part B

Test

database

Test

Data

Part B

Call Home

Screening / Failure

Analysis Centers

40% of customers

File Incident Ticket

RMA

Receipt

0.5%

99.5%

SAP

Repair-1

(Component)

Repair-2

(Component)

Repair Centers /

sub-tiers

Repair/

FA data

Home Grown Repository (OEM)

Supplier Repository

ERP / CRM software (OEM)

Legend

Other

Reports

Test & Repair

SQE FA

Reports

Internal

RMA

(System)

Engineering

Feedback

Supplier

Root Cause

FA data

Repair/

FA data

Partial Data

Fragmentation

> 15 Data Sources

Typical Scenario - Data,

Data Everywhere

Customers Oracle

RCA

3PL-1

(Module)

3PL-2

(Module)

3PL-1

(System)

Oracle

Quality

Corporate

Reliability

Reports

IFS /

Stars /

Tars

XML

(7C6)

Pass/Fail Data

Desktop

Database

Dept.

Database

Test Data

Part B

Test

database

Test

Data

Part B

Call Home

Screening / Failure

Analysis Centers

40% of customers

File Incident Ticket

RMA

Receipt

0.5%

99.5%

SAP

Repair-1

(Component)

Repair-2

(Component)

Repair Centers /

sub-tiers

Repair/

FA data

Home Grown Repository (OEM)

Supplier Repository

ERP / CRM software (OEM)

Legend

Other

Reports

Test & Repair

SQE FA

Reports

Internal

RMA

(System)

Engineering

Feedback

Supplier

Root Cause

FA data

Repair/

FA data

Partial Data

Latency

30 days

Typical Scenario - Data,

Data Everywhere

Building an Early Warning System

To Expedite RCA

Leading Risk Indicators

What happened ?

Why ?

-What is the root cause

-Is it a Design, Process or Supplier Issue?

-How do I prevent it from happening again

Demo

Using Data To Accelerate RCA

Cultivate holistic data strategy

Invest in Early Warning to accelerate

RCA

Empower intellectual resources to

make better decisions, sooner

Contact Information

For more information visit:

www.sigmaquest.com

Contact Information:

Al Alaverdi

408-524-3181

al.alaverdi@sigmaquest.com

Question & Answer

Microsoft Root Cause Analysis Webinar Introduction V2 987 RCA By Ops A La Carte

Navigation menu

Versions of this User Manual:

Views

Navigation