Levels of Failure:
Apply Hazard and Operability Sutdy, Hazards Analysis, Fault Tree Analysis, Failure models and effects analysis, and Ishikawa diagrams
Failure analysis is post-failure forensics
Hazard analysis is pre-failure/preventative analysis to avoid potential disasters - expect the unexpected
Types of Data Analysis
Diagnosis
Prognosis
Levels of Failure
One - physical flaws (mostly mechanical engineers)
Two - process (hits us the most out of the 3)
Three - perspective or attitude
PHysical flaws
Process
Perspective or Attitude
Ask Five Whys
Why? Battery dead
Why? Alternator not functioning
Why? Alternator belt broken
Why? Alternator belt beyond useful service life
Why? Vehicle not maintained according to recommendation service schedule
Root Cause Analysis (RCA)
Method to identify root causes of faults or problems in four steps
Toyota Way (Toyoda?)
14 Principles in four sections
Base management decisions on long-term philsophy at expense of short-term financial goal
Standardize task and process are foundaation for continuous improvement (kaizen) and employee empowerment
Build culture to fix problems and get quality right the first time
heijunka kaizen genchi genbutsu nemawashi hansei
A3 problem solving
A3 paper size is size of the table
Tabular form, identify things like name, leader, stakeholders, etc.
Specify problem, root cause, breakdown, etc.
Helps solve problems
Fault Tree ANalysis(FTA) (top down)
a bunch of symbols
FMEA - bottom up
Walter A. Shewhart (Bell Labs)
Father of control chart and statistical process control to identify outliers and did further
W. Edwards Deming
Took Shewhart's process and introduced it to lecture he gave in Japan
"If you can't describe what you are doing as a process, you don't know what you're doing."
"It is not enoguh to do your bets; you must know what to do , and then do you rbets."
When Japanese find something good, they will make it the best.
Kaoru Ishikawa
Plan Do Check Act (Deming) into six steps
Seven basic tools of quality
scatter diagram
Standards not the ultimate source
Ishikawa (Fishbone) Disagnostic Diagram
Zipf's Law (1935)
harvard uni prof, linguist
Mathematical model, log-log is straigh tline
Discrete Pareto distribution
Vilfredo Pareto
Italian civil engineer
Pareto distriution which is power law probability distribution
80/20 rule (80% of land owned by 20% of population)
Pareto chart is special type of histogram used to view causes of a problem in order of severity from largest o smallest
Six sigma
registered trademark and service mark fo motorola
Six Sigma process improvement efforts, defects per million opporunities
Specification lmits (USL and LSL) are distance 6sig from mean
Even if mean move lft or right by 1.5sig, there is good safety cushion
Methodloogies (six sigam)
DMAIC improve existing business processes
DMADV creat new product or process designs
Capability Maturity Model Integration (important for sofwater engineer)
Carngeie Mellon University
5 levels
5 - optimizing focus on process imporvoement
4- quantitiavile managed process mesausred and controlled
3- defined process characterized for the organization and is proactive
2-managed processes characterized for projects often reactive
1- intiial process unpredictable poorly controoleed
Dealing with hazards (5 steps)
NEC (national electrical code) prevents fire
National Electrical Safety Code is about user safety (ANSI Standard C2)
Motor Vehicle Safety Standards (eg antilock brakes, seatbelt, airbag)
Regrettable inventions
Did not know about lead in gasoline, CFCs in refrigerants, propellants in aerosol applications
Jeff Bezos
Idea -> When I'm 80 will I regret not doing it -> No let it go, yes do it
He's no longer WallStreet trader
Circular Economy
we want to reuse
Hazard Avoidance
Use guards, sensors, interlocks, and other metchanisms to distnace user form hazard
Redundancy, backups, etc.
Safety signs as opposed to warning signs
SecDevOps (Secur DevOpment) best practices
plan code build test release deploy operate monitor repeat
Chaos Engineering
Experimenting on software system in production to build confident in system's capabilities to withstand turbulent and unexpected conditions
Simian Army is suite of tools by Netflix to test reliability security etc of its AWS infrastructure
Chaos Monkey 2011 to test resilience of IT infrastructure
10-18 monkey detetct problems with local/internationalization (l10n-i18n)
Why monkey? Because they usually have smiley face and han dout banana on street to promote AWS
Geohazards
Zoombombing
OSHA