Dictionary ZQS

You might also like

Download as doc
Download as doc
You are on page 1of 16

ANOVA ANalysis Of VAriance An ANOVA is an analysis of the variation present in

an experiment. It is a test of the hypothesis that the variation in an


experiment is no greater than that due to normal variation (random noise)
of individuals' characteristics and error in their measurement. (Ratio of
explained to unexplained variation.
Note-1: The output of ANOVA table is the F-test.
Note-2: We are comparing MEANS (because we assume variance to be
equal)
Also see ANOM (analysis of means)
Most accurate method for quantifying reproducibility (technician variance) & repeatability (error variance)
Benchmarking Steps of Benchmarking
1. Determine current practices and processes to be benchmarked. Understand your and others processes.
2. Identify best practices and areas for improvement. Measure your performance & define leaders in best
practices. (do not just copy someone’s process)
3. Gather data. Analyze best practices.
4. Repeat the cycle.
Box & Whisker
or Box Plot

Outliers are values that


are more than 1-1/2 box
lengths away from upper
or lower edge of box
50th percentile is median, 25th percentile is “median” of lower ½
BPM Business Process Management: is a method of efficiently aligning an organization with the wants and needs of
clients. It is a holistic management approach that promotes business effectiveness and efficiency while striving
for innovation, flexibility and integration with technology. As organizations strive for attainment of their
objectives, BPM attempts to continuously improve processes - the process to define, measure and improve your
processes – a ‘process optimization' process.
Chart CPM Critical Path Method (See PERT) ACTIVITY oriented Note: can be more vigorous than PERT
Chart PERT Program Evaluation & Review Technique ( see CPM) EVENT oriented)
Chart Scatter – r Explained variation
Linear Total variation (NOTE: as explained variation increases the correlation between x & y increases)
Regression • You choose values of x to see how it affects y (x = independent variable, y = dependent variable)
• Simple linear regression equation is y=ax+b
y=dependent variable, x=independent variable, a=slope, b=intersect (where avg. value of x intersects average value of y)
Note: If there is a correlation it DOES NOT ALWAYS mean that one caused the other (I,e. shark attacks and ice cream sales)
r calculated means strong correleation if r = + or -1 (45º slope) and no correlation (no slope) if r = 0
COF Coefficient of Variation =Std-dev/Mean (expressed as %)
Control Chart
Decision Tree
Control Chart • Attribute charts are useful for both machine- and people-based processes. Data for them is often
Defined readily available and they are easily understood. It can thus be easier to start with these, then move on
ATTRIBUTE to Variables charts for more detailed analysis.
CHARTS
• u- and c-charts give a measure of DEFECTS (can be reworked or does NOT render product
unusable). You cannot count the non-occurrences. Failures either happen or they don’t.
• u-chart, VARYING sample size (%) the defects within the unit must be independent of one another,
for example, 'component failures on a printed circuit board, or defects per square meter of cloth'.
• c-chart FIXED sample size is a useful alternative to a u-chart when there are a lot of possible defects
on a unit, but there is only a small chance of any one defect occurring. For example, 'flaws in a roll of
material, or dents per panel'.
• p- and np-charts give a measure of DEFECTIVES (unit failure or cannot be repaired). You can count
the occurrences and non-occurrences. Failures may happen more than once.
• p-chart, VARYING sample size (%).

• np-chart. FIXED sample size


Control Chart • Variables charts are useful for machine-based processes, for example in measuring tool wear.
Defined • Variables charts are more sensitive to change than Attributes charts, but can be more difficult both in the
VARIABLE
identification of what to measure and also in the actual measurement.
CHARTS
• XmR (aka: I-MR) Individuals & Moving Range.
• Only use an Individuals (X) chart when few measurements are available (e.g. when they are infrequent or
are particularly costly. Data must be normally distributed. These charts are less sensitive to change than
the Averages (X-bar) chart. They can also have an unusual (non-normal) distribution, in which case the
Control Chart is not a suitable tool (use a Line Graph instead).
• Moving Average Chart: Use to smooth data and estimate future values. Based on X-bar and S chart. Best
when mean is stable, poor when process is trending. (Note: 1st subgroup =5 then avg. of 1st group&7, avg.
of 7&8, avg. of 8&9….)
• The Standard Deviation (s) chart is used to monitor variation. May be used instead of the Range chart.
Used for large subgroups (>25).

X-bar & Range chart should have 100 individual data points at a minimum to calculate control limits.
Control Chart
formulas
Attributes
c-chart

Control Chart
formulas
Attributes
np-chart

Control Chart
formulas
Attributes
p-chart
Control Chart Note: This is
formulas a “running”
Attributes calculation
u-chart
Control Chart X-bar & R chart n A2 d2 D3 D4 E2
formulas
VARIABLES
for X-bar: UCLx = x + A2 ( R ) LCLx = x − A2 ( R ) [A2 for n=5 is 0.577] 2
3
1.88
1.023
1.128
1.693
--
--
3.267
2.574
2.66
1.77
4 0.729 2.059 -- 2.282 1.46
for Range: UCLR = D4 ( R ) [D4 for n=5 is 2.114] Note: Zero is LCLR for 5 0.577 2.326 -- 2.114 1.29
6 0.483 2.534 -- 2.004 1.18
n=5 7 0.419 2.704 0.076 1.924 1.11
8 0.373 2.847 0.136 1.864 1.05
XmR chart 9 0.337 2.97 0.184 1.816 1.01
10 0.308 3.078 0.223 1.777 0.98
mR mR
for X-bar: UCLX = X + 3 LCL X = X − 3 [d2 for n=2 is
d2 d2
1.128]
for Range: UCLR = D4 (m R ) [D4 for n=2 is 3.267]

Alternate XmR X-bar

for X-bar: UCL X = X + E 2 (m R ) LCLX = X − E2 (m R ) [E2 for n=2 is 2.66]


Control Chart 1. Process must be stable
Pre-Control 2. Process spread must be less than or equal to specification
spread (Cp ≥ 1.0)
3. Ideal is 25 checks until reset is required, if reset is required
before 25 checks increase sample frequency if no reset
required in 25 checks relax sample frequency
Rules Setup
• Ok to run if 5 pieces in a row are inside limits
Rules running: Sample 2 pieces in a row
• If 1st or 2nd piece is out of Spec limits STOP
• If 1st piece is inside P/C limits run & don’t measure 2nd piece
• If 1st piece is not in P/C limits check 2nd piece
• If 2nd piece is not in P/C limits STOP otherwise run
NOTE: Uses Attribute data (pass/fail)
Control Chart WECO (Western Electric COmpany rules) General rules for detecting out of control or non-random situations
Rules
Any Point Above +3 Sigma
--------------------------------------------- +3 LIMIT
2 Out of the Last 3 Points Above +2 Sigma
--------------------------------------------- +2 LIMIT
4 Out of the Last 5 Points Above +1 Sigma
--------------------------------------------- +1 LIMIT
8 Consecutive Points on This Side of Control Line
=================================== CENTER LINE
8 Consecutive Points on This Side of Control Line
--------------------------------------------- -1 LIMIT
4 Out of the Last 5 Points Below - 1 Sigma
---------------------------------------------- -2 LIMIT
2 Out of the Last 3 Points Below -2 Sigma
--------------------------------------------- -3 LIMIT
Any Point Below -3 Sigma
TREND & SHIFT rules are for X-bar and Range on X &R charts and X-bar on XmR charts ONLY
Trend Rules: 7 in a row trending up or down. 14 in a row alternating up and down
Shift: 7 or more above or below average
Odds of 7 in a row are (1/2)7 = 1/128 but since it could go above or below (50% chance) = 1/64
Control Plan Document describing CTQ characteristics, the critical X’s or Y’s of the part or process. Every completed Six
Sigma project should have not only a control chart (if applicable), but a control plan. This ensures that the
process doesn't revert to the way it previously operated. This is a living document and is rarely closed.
Cost Of Quality See Quality Costs
Cpk, Ppk, Cp, USL − LSL USL − LSL USL − X X − LSL
Cp _ or _ Pp = S = Cpk _ or _ Ppk = or ( = min_ result )
Pp, Formulas 6σ 6 •Cp
3σ 3σ
(see also
Z min USL −x −bar x −bar −LSL
“Performance Ppk = [
3
] Z min =
3
or
3
( =min_ result )

Ratio”)
Cpk, Ppk, Cp, Cp & Pp = Capability
Pp, values Cpk & Ppk = Acceptability
Distribution shape = Normality (normal is bell shaped).
Cp and Pp – <1.00 incapable
1.00 to 1.33 capable with tight control
>1.33 capable
Histogram – bell shaped
one peak
can be slightly skewed
Cpk or Ppk – <1.33 not acceptable
1.33 to 1.50 acceptable if process stays centered
>1.50 acceptable with wiggle room
Note: Cp is generally larger than Cpk
Note: For non-normal data use some form of Data Transformation
CTQ Critical To Quality
Design Set Based Design: Incorporated various functions & concepts, Begins with broad based possibilities and
narrows them down to a few alternatives and then to a final solution.
Systematic: Step-by-step approach
DATA Types Attribute Data
(see also – Discrete / qualitative / categorical / attribute / counted
“Scales of • Integers (WHOLE numbers: 1, 2, 0, -2) cannot be divided any smaller (i.e. cannot have 2.4 dents in car).
Measurement”) • One or the other (hot cold, yes no…)
– Counted
• How many defects
• How often do they occur
• What kind or category
Variable Data
– Continuous / quantitative
• Any real number (any numeric value and can be meaningfully subdivided into finer increments).
• Information that can be measured on a continuum or scale.
– Measured
• How long did it take
• How far does it reach
• How much does it weigh
Diagram
Activity
Network
Diagram Options that meet "MUST" criteria
Affinity Refurbish existing Purchase new
Wants* Weight ** machine machine Farm job out
Least Cost 6 8 48*** 4 24 10 60
Reliability 10 8 80 8.5 85 4.5 45
New features 9 5 45 10 90 3 27
No new training 4 9 36 5 20 8 32
161 219 164

* Wants criteria are developed to judge the options


** At least one Weight gets a value of 10, the rest are compared to this (as voted on)
*** = Weight x value of each category (48 = 6 x 8)
Note: Use "MUST" criteria when establishing options (i.e, must improve quality and output)

Diagram Matrix
(also PUGH) * Displays relationships between
Tasks and people
Tasks and tasks
* Analyze the correlations between two ideas
*
Shows importance of each connecting point
relative to every other correlation
* Displays the relationship between necessary
tasks and people or other tasks, often shows
responsibility for tasks

Diagram PDPC Process Decision Program Chart


Chart * Analyze where problems (or
failures) can occur
* Design to prepare for contingency
plans
* Helps develop specific actions to
be taken to prevent problems from
occuring (or if they occur, to
minimize their impact)
Based on PDCA (Plan Do Check
*
Act) model

Diagram Tree Cost within budget


Returnable if defective
(CTQ diagram) Affordability
Long lasting
wear repeatably

Shock absorbing soles


A shoe that's worth Comfortable Soft lining
the price fit Lightweight
Room in toe area

Color
Attractive Modern
style Low heel
Leather

Diagram Venn

Distribution The discrete probability distribution of the number of successes in


Binomial a sequence of n independent yes/no experiments, each of which
yields success with probability p. Example-1: There is a test where
students will guess at the answers with 1- Each question has 4 multiple choice
questions (only 2 states, right or wrong, trials are independent and mutually
exclusive (include all possible choices), probability on each trial is the same) 2 -
Each test has 25 questions ( Fixed number of trials). Example-2 Output of a
process has x defectives (characteristics).
Distribution Joint distribution of two variables (3D Graph)
Bivariate
Distribution A continuous probability distribution estimating the variation (standard deviation)
Chi-Square of 2 populations. Tests for “goodness of fit” Example: Sample of 10 drawn from normal
universe whose variance = 25. The variance of the sample is 35.Question: is there a statistical
significance between the two numbers (25 & 35) at a 95% confidence level? (H0: variance of observed =
variance of expected). Test for independence & goodness of fit is always right tailed test.
Distribution TIME between successive events. In probability theory and statistics, the
Exponential exponential distributions are a class of continuous probability
distributions. An exponential distribution arises naturally when modeling
the time between independent events that happen at a constant average
rate. i.e. MTBF
Distribution F A continuous probability distribution used to calculate the ratios of variances (sample variances are assumed to
be equal) of normally (assumed) distributed populations (is sample-1 from a population with the same standard
deviation, within a specified confidence level the same as sample-2). Example: Company A claims it is 99% confident that
its machine has less variation than company C’s machine. (Sample size need NOT be the same and we must know standard deviation (or
variance) of both machines.
Note: The F distribution is the ratio of two chi-square distributions, also it is used in ANOVA
Also known as Snedecor’s distribution.
Distribution A discrete probability distribution that describes the number of successes in a
Hypergeometric sequence of n draws from a finite population without replacement. Defects Shape is similar to
(characteristics) are known for the population. Sample size > 10% of N, 2 Binomial/Poisson
states, red/black, fixed number of trials. Example: drawing 5 cards from a deck, what is the distribution
probability for “x” (1, 2, 3, 4, or 5) red cards being drawn?
Distribution A continuous normally distributed probability distribution with a mean of zero and a standard deviation of one.
Normal, Z- Bell shaped curve is due to central limit theorem. Large sample (>30 if <30 then use T distribution). Example:
table Manufacturer tested a large sample of one of their parts and found the length averaged 10.000 inches with a standard deviation of 0.050
inch. Q1 – How many parts are expected to be larger than 10.122 inches?
(also called the Gaussian distribution)
Distribution A discrete probability distribution that expresses the
Poisson probability of a number of events occurring in a fixed
period of time if these events occur with a known average
rate (population average and # of occurrences are known)
and independently of the time since the last event. Example:
The average number of people on hold during working hours for service
calls is 4 during working hours on any give day. Q1 - What is the
probability that there is nobody on hold? Q2 - What is the probability that
there are 4 or fewer people on hold?
Note: Used for number of occurrences (DEFECTS) of some event (for time between events see exponential
distribution). Is used with RATES, FREQUENCIES, and FLOW. Mean = variance
Distribution A continuous probability distribution estimating the mean of a normally distributed population when the
Students t sample size is small (<30 if >30 then use Z or Normal distribution). Used to develop a confidence interval for
μ. Compare "means" of two data sets (we assume std-dev of the 2 populations is the same) Example: Calls are to be
answered within 15 seconds at a call center. The claim is that with 99% confidence the average response time is less than 17 seconds.
Random sample of 25 calls reveals that x-bar= 13.5 sec. and s = 0.95sec. Is the claim correct?
Distribution T The basic difference is that the paired t requires the 2 sample sets to be dependent where the student's t sample
Test – paired & sets must be independent of each other. The sample sizes in the paired t test are equal while in the student's t test
student they don't have to be equal. The paired t tests the difference between 2 population means where each data point
in the first sample is paired with a data point in the second sample (the data sets are dependent on each other
such as before and after process change). The student's t test the difference between 2 means from 2
independent populations, in other words, they are unrelated.
Distribution Most common in reliability & statistical applications, used to model time to fail, time to repair, material strength
Weibull (failure). Scale parameter (determines width of distribution) β. If β = 1 weibull = exponential, if β = 2 weibull =
Rayleigh, if β = 3-4 weibull approximates Normal.
DMADOV Define Measure Analyze Design Optimize Verify
DMADV Define the project and the goals of the design activity. What is being designed? Why? Use QFD or Analytic
Hierarchical Process to assure that the goals are consistent with customer demands and enterprise strategy.
Measure the current state and determine Critical to Stakeholder metrics. Translate customer requirements into
project goals.
Analyze the root causes and the options available for meeting the goals. Determine the performance of similar
best-in-class designs.
Design the new product, service or process. Use predictive models, simulation, prototypes, pilot runs, etc. to
validate the design concept's effectiveness in meeting goals.
Verify the design's effectiveness in the real-world.
(Associated with DFSS Design For Six Sigma)
DMAIC Define Measure Analyze Improve Control
tollgates

DMAIC vs.
DMADV
DOE terms Blocking: The arrangement of experimental units into groups (blocks) that are similar to one another.
Confounding: Aliasing (factor-A, factor-B, factor-C, AB+C = C is aliased with the interaction AB) (confused)
Covariance: a measure of how much two random variables vary together (are correlated) (Covariate).
Collinear: variables that are linear combinations of one another (high correlation among the variables).
Orthogonal: Balanced, no confounding, equal no. of data points under each level of each factor (full factorial)
(Latin Square: equal no. of rows and columns)
Randomization: Conducting experiment in no given order.
Replication: Performing the same treatment combination more than
Resolution: If resolution = 3 then 1 & 2 interaction, if 4 then 1 & 3 and 2 & 2…. (FIVE FINGER RULE)
Input Variables: Independent and predictor variables (NOTE: may be attribute data)
Output Variables: Dependent variables
DF: degrees of freedom (n-1) is required for each factor where n is the number of levels for the factor (i.e. 2
factors F1=3 levels & F2 = 4 levels so DF= (n1-1)(n2-2) = (2)(3) = 6

There should be no more than 2-4 factors in a experiment. If more try to screen out some of the list.
Items that a statistician may be lacking: 1-unwarranted assumptions of ht process 2-undesirable combinations of the factors 3-violation of
known laws of physics 4-too large or small design size 5-inappropriate confounding 6-imprcise measurement 7-unaccaptable prediction
error 8-undesirable run order
DPMO / DPMO - Defects per Million Opportunities - Credit received for meeting SOME of the requirements
DPMU DPMU - Defects Per Million Units - Credit given only when ALL requirements are met.
(Determining DETERMINE BASELINE SIGMA
baseline sigma) Example: 20 defects found in 200 units with 3 possible defects per unit
see also TDPU
DPMO = Defects/Units x Opportunities = 0.0333
0.0333 x 106 = 33,333 is between 3.3 & 3.4 sigma*

DPMU = Defects/Units = 20/200 = 0.1


0.1 x 106 = 100,000 is between 2.7 & 2.8 sigma*
*NOTE: Sigma values as shown in Six Sigma Failure Rate table
DPU / DPO Defects Per Unit Defects Per Opportunity
Example: 15 defects found in 250 units with 5 possible defects per unit.
Defects/Units = 15/250 = 0.06 DPU -We would expect to find 0.06 defects per unit.
DPO = Defects/Units x Opportunities = 15/250 x 5 = 0.012
0.012 in Z table = 2.26, 2.26 + 1.5 shift = 3.76 sigma level
(NOTE: 0.012 x 106 = 12000 DPMO)
ERROR Types • Type I Error – α - (alpha) - Null hypothesis (are =) is rejected when it should not have been (Setting
alpha to .05 vs .01 means more risk of type I error) if is rejected and should have been then 1-β
• Type II Error – β - (beta) - Null hypothesis (are =) is not rejected but should have been (Setting alpha
to .05 vs .01 means less risk of type II error) if is not rejected and should not have been then 1-α
EVOP In EVOP Evolutionary operations experimental designs and improvements are introduced, while an ongoing
full-scale manufacturing process continues to produce satisfactory results. The idea is that process improvement
should not interrupt production. Testing & data collection is typically performed by operators.
Failure Project 5 top reasons for project failure:
1. Poor team dynamics
2. Scope creep
3. Poor or no causation (the relation of cause to effect, failure to answer why a problem exists)
4. Poor or no champion involvement
5. Failure to obtain stakeholder buy-in
Five From Michael Porter of Harvard Business School
Competitive 1. The threat of new entrants
Forces 2. The power of customers
3. The power of suppliers
4. Substitute products or services
5. Industry rivalry
Five S’s Sort, Store, Shine, Standardize, Sustain (safety is considered 6th S)
FMEA Failure Mode Effects Analysis
F-failure-DEFECTS
M-mode-WHAT GOES WRONG
E-effects-CONSEQUENCES
A-analysis-RISK ASSESSMENT

FMECA Failure Mode, Effects, and Criticality Analysis


Used to chart the probability of failure modes against the severity of their consequences (RPN no).
Guru’s of Deming
Quality • PDSA
• 14 step approach to improvement
• 7 deadly diseases
• Chain reaction “improve quality, decrease costs, improve productivity, capture the market with better
quality and price, stay in business, and provide jobs”
Ishikawa
• Quality circle
• CWQC (Company Wide Quality Control)
• Fishbone diagram
Juran
• Trilogy
• Proponent of improvement & breakthrough projects
Shewart (Western Electric)
• Control charts
• PDCA
Taguchi
• Robust DOE (use experimentation & testing to establish functional limits)
• Quality engineering pioneer
Crosby
• 14 step approach to improvement
• 4 absolutes (conformance to requirements, prevention, zero defect standards, quality measurement is $)
Histogram & Determining Cell Intervals
Frequency No. of
Distribution data No. of
points cells
Under 50 5-7
50 - 99 6 - 10
100 - 250 7 - 12
> 250 12 - 20
35 is divisable
Example: 70 data points with 7 by 7 so use 7 cells
highest value = 83 83 (49-53, 54-58…)
lowest value = 49 -49
= 34
(for inclusion) +1
= 35

Histogram will NOT show patterns in data such as Trends, Shifts, 8 Consecutive Points above or below mean, etc.)
House Of QFD (Quality Function Development)
Quality (HOQ) Roof Interaction & Competitor Correlation Key Correlation matrix
• P = positive & PP = strong positive HOW'S

• N = negative & NN = strong negative


Value WHAT'S
requirements

Action Notes
Importance

Competitor
Customer

Rating

Relationship Matrix

IDOV Identify Design Optimize Verify


Ranking of HOWS
Interval • Used with inferential statistics to develop a confidence interval
Estimation • The interval within which it is believed (with a given degree of confidence) the population parameter
(confidence) lies
Confidence + Significance = 100 % Example: Confidence = 95%, Significance (alpha α) = 5%
Kaizen Continuous improvement philosophy.
By improving the standardized activities and processes, Kaizen aims to eliminate waste
Lean 5 steps 1. Specify Value
2. Map the value stream
3. Create flow
4. Pull
5. Pursue perfection
Little’s Law Lead Time = WIP (units) / ACR (units per time period)
Management Functional Management (vertically within function):
Types • Budget adherence
• Command and communication throughout the organization
Process Management (horizontally between departments): best for customers
• Increases effectiveness and efficiency
• Increases customer satisfaction
• Enables achievement of business objectives
Math sequence Please Excuse My Dear Aunt Sally
of execution Parentheses
Exponents
Multiply
Divide
Add
Subtract
Measurement Use X-bar & R chart to show measurement system variation. Example below 5 parts measured 3 times (same
Systems location on part) and measured by 3 people.
Evaluation The range of measurements within the same piece establishes the control limits and since there should be no
variation on the piece being measured the control limits are generated by the range of the measurement
variation.

The control limits represent the range (variation) of the measurement system.
Measurement • BIAS - Difference between the output of the measurement system and the true value, i.e., 1.00000”
systems gage block is measured at 1.0001” to 1.0005” so bias is .00004 (absolute value of: average measured –
variation average of known).
• LINEARITY - is the difference in the bias values through the expected operating range of the
Gauge.
• REPEATABILITY - is the variation in measurements observed over several measurements made
under the same conditions.
• REPRODUCIBILITY – Average variation in measurement due to factors other than the
measurement system or instrument (such as operators, different gages, environmental changes, etc.)
• Stability is the consistency of performance over time.
• Accuracy and sensitivity must be assured before R&R can be performed
MTBF MTBF Mean Time Between Failure (repairable)
3 units evaluated over 1000 days
Unit No. Failures
1 2 Failures = 10 = 3.33 (the average unit will need repair
2 4 units 3 3.33 times in 1000 days
3 4
Total 10 Total time period = 1000 days = 300.3 (unit will last an average
Mean No. of failures 3.33 of 300.3 days before
repair is needed)

Failure rate = 1/MTBF = 1/300.3 = .00333 (this means there is a .33% chance that a single
unit will fail on a single day)
Exponential Distribution
MTTF MTTF Mean Time To Failure (non-repairable)
3 units evaluated over 1000 days
Fuse Life (hours)
1 1050 Life = 3400 = 1133.3 hours (the fuses have an average
2 1200 units 3 life of 1133.3 hours
3 1150
Total 3400 Hazard rate = 1/MTTF = 1/1133.3 = .00088 (Example: it is specified that fuses
should have a reliability of 99%
after 100 hours)
Time x hazard rate = probability of failure
100 x .00088 = .088 or 8.8% chance of failure after 100 hours
Exponential Distribution
MUDA The Japanese term for waste.
Multi-Vari • Dramatize variation within a piece (or process, 1 2 3 4
Analysis i.e., oven temperature at various locations in oven) Example: part measured across face at edge
i.e, assists in the breakdown of components of variation
• Dramatize variation from piece to piece
• Track time related changes 4 2,3

• Helps identify areas to look for excessive 3


4

variation & also identifies areas not to look for 3


2
excessive variation. 2
Variation = piece (positional), piece-to-piece (cyclical), 1 1,4
1
time (temporal) FIG-1 FIG-2 FIG-3
Part is tapered Center is thicker Part is getting larger
Null • Always has some form of equal sign, =, ≤, ≥
Hypothesis • If test is for being equal to i.e. H1 = H2 then 2 tailed test
• If test is for being less (or greater) than or equal to i.e. H1 ≤ H2 then 1 tailed test
• Null Hypothesis assumes any variation in results is due to chance (common cause) alone.
• Alternative Hypothesis is what we expect to find (assignable cause from one of 6 M’s).
• Null always says that there IS NOT or MAY NOT be a difference.
• Null Hypothesis may be REJECTED but is NEVER ACCEPTED, we only FAIL to REJECT it.

Example above: 95% area is “fail to reject null” area so 5% of the time we are willing take a chance (type I
error) that we reject the null when the variation was due to chance alone.
PDCA / PDSA Plan, Do, Check, Act cycle. Shewhart Plan, Do, Study, Act cycle. Deming
PLAN: plan ahead for change. Analyze and predict the results.
DO: execute the plan, taking small steps in controlled circumstances (i.e. pilot program).
CHECK / STUDY: check or study, the results.
ACT: take action to standardize or improve the process (i.e. act on resulst of pilot program).
Performance 6σ
Ratio CR CR =
(USL − LSL)
PEST The acronym for "Political, Economic, Social, and Technological environmental issues that an organization
faces (internal)
Power of The power of a statistical test is the probability that the test will reject a false null hypothesis (that it will not
Statistical test make a Type II error). As power increases, the chances of a Type II error decrease. The probability of a Type II
error is referred to as the false negative rate (β). Therefore power is equal to 1 − β.
Probability Probability definition:
Theory The probability of an event is a number between 0 (cannot occur) and 1 (will occur)
0 < P(A) < 1

INDEPENDENT: Events do not depend on each other (sampling with replacement)


Example: Urn contains 5 black & 3 white balls. If 2 balls are drawn (WITH replacement) what is the probability both are black?
• Let A be the event that 1st ball is black and B be event that 2nd ball is black.
• Then P(A) = 5/8 and P(B) = 5/8
• P(A and B) = P(A) x P(B) = 5/8 x 5/8 = 25/64

MUTUALLY EXCLUSIVE: events cannot occur together (sampling without replacement such as drawing a
ace OR a king from a deck of cards)
EXAMPLE:
• A is event of drawing a ace and B is event of drawing a king
• P(A) = 4/52 = 1/13 & P(B) = 4/52 = 1/13
• P(A or B) = P(A) + P(B) = 1/13 + 1/13 = 2/13 probability of drawing either a ace or king in a single draw
CONDITIONAL PROBABILITY: the probability of B occurring given event A has occurred. Example: Train
arriving on time = A (0.84), leaving on time = B (0.86), arriving AND leaving on time = B+A (0.8) so p of B if A has occurred
p ( A + B)
p( B / A) = 0.8/0.84 = 0.9524
p ( A)
Process 1. Name the process (name is critical i.e. “getting a car” is to vague, use buying a new car, buying a used car, leasing a car)
Mapping 2. Identify start and stop points (boundaries) (scope and scope creep)
3. Identify the output of the process (unqualified noun)
4. Identify the customer(s) of the process
5. Identify the supplier(s) of the process
6. Identify the inputs of the process
7. Indicate the 4-7 highest level steps in the process as the exist today.
See SIPOC
Project or A collection of documents that provides purpose and motivation for an improvement team to do its work.
Team Charter SERVES THREE FUNCTIONS:
• Establish focus for the team
• Motivate the teams behavior
• Motivate emotion
Six elements of a good charter:
1. Vibrant business case (what the project does and how it impacts the business objectives, answers
question why is the project worth doing and what are consequences of not doing project)
2. Problem statement (be specific and measurable, indicate how long problem has existed, describe impact
of problem, describe gap between current and desired state, and use neutral terms)
3. Goals and Objectives (may not have all data in the beginning)
4. Project Scope (defined boundaries what is included and what is not included)
5. Realistic milestones & deliverables
6. Clearly defined roles and responsibilities
Forming, storming, norming and performing (growing a team)
Project Planning, Scheduling, Controlling
Management Project is a series of activities and tasks with a specified objective, starting and ending dates, and necessary
resources. Resources consumed by the project include time, money, people, and equipment.
PUGH Analysis See “Diagram Matrix” - based on the use of matrix based process to refine conceptual design
QFD Quality Function Development (see House of Quality)
Quality Costs • Prevention Costs: activities designed to prevent poor quality (repair, capability studies, design review,
field testing, training, fixturing, surveys, evaluating vendors)
• Appraisal Costs: costs associated with measuring, evaluating, auditing (calibration, inspection,
laboratory testing, test equipment maintenance)
• Failure Costs: costs resulting from poor quality
1. Internal: Prior to delivery or shipment of product or service
2. External: After delivery or shipment of product or service.
Hidden Failure Costs: - Engineering Time – Management Time – Downtime – Increased Inventory –
Decreased Capacity – Lost Opportunity
QxA=E Q = quality of solutions (scale of 1-10) A = Acceptance of solutions E = Excellence of the results
E must = 60, in order for this to happen A cannot be ignored in favor of Q (your solutions must be accepted)
Regression See Chart scatter
Analysis
ROI & ROA Net _ income Net _ income
Return On Investment ROI = Return On Assets =
Investment Total _ Assets
Roles & Titles Sponsor Senior executive who sponsors the overall Six Sigma Initiative.
Six Sigma Senior-level executive who is responsible for implementing Six Sigma within the
Leader business.
Highly experienced and successful Black Belt who has managed several projects
and is an expert in Six Sigma methods/tools. Responsible for
coaching/mentoring/training Black Belts and for helping the Six Sigma leader and
Master Black Belt Champions keep the initiative on track.
Full-time professional who acts as a team leader on Six Sigma projects. Typically
has four to five weeks of classroom training in methods, statistical tools, and
Black Belt (sometimes) team skills.
Part-time professional who participates on a Black Belt project team or leads
smaller projects. Typically has two weeks of classroom training in methods and
Green Belt basic statistical tools.
Middle- or senior-level executive who sponsors a specific Six Sigma project,
Champion ensuring that resources are available and cross-functional issues are resolved.
Professional responsible for the business process that is the target of a Six Sigma
Process Owner project. (also Champions & Sponsors)
Professional who has general awareness of Six Sigma (through no formal training)
Team Member and who brings relevant experience or expertise to a particular project.
RPN Risk Priority Number Example – flat tire on car:
-Function: hold air -Occurrence: 2 (P) = probability of occurrence (1=small chance)*
-Failure mode: does not hold air -Prevention: maintenance
-Effects of failure: stranded -Detection control: warning light
-Severity: 8 (S) (10=severe)* -Detection: 1 (D) = effectiveness of detection control (1=effective)*
-Cause: bent rim
RPN = PxSxD (*scale of 1-10) this example = 2x8x1 = RPN = 16
Scales of Nominal scale
Measurement • Identifies category to which a unit belongs (i.e. Sex: 0-male, 1=female)
• Assigned numbers have no value
Ordinal scale
• Units are ranked (i.e. a scale of 1 to 10)
• Only the order of numbers is meaningful
Interval and ratio scales
• Actual Measurement where numbers have a value (i.e. height in inches)
• Scales dependent on method of measurement not property being measured (i.e. temperature measured
relative to absolute zero)
Scree Plot In this example: Sum of Square s Scre e Plot
Factor Delta SS
• C, G, A, & F are C 29.1 106 120
100
G 27.3 93
significant A 25.0 78
80
SS

60
F 23.0 66
• D, E, & B are noise. D 13.0 21
40
20
E 6.3 5 0
B 4.0 2 C G A F D E B
Factors
Note: SS= ( value)²
Seven n (factors)
The seven basic Management Tools: The seven basic tools for TQM are:
Management 1 Interrelationship digram 1Cause and effect diagrams
Tools & basic 2 Affinity diagram 2 Flow charts/process flow diagrams
tools for TQM 3 Tree diagram 3 Pareto charts
4 Prioritization matrices 4 Run charts
5 Matrix diagram 5 Histograms
6 PDPC process Decision Program Chart 6 Scatter diagrams
7 Activity network diagram 7 Control charts

SI The International System of Units. Meter, Kilogram, Second, Ampere, Kelvin, Mole, Candela
Sigma Value
(ppm defects) ppm Defects ppm Defects
Sigma Percent With 1.5 σ Sigma Percent Without 1.5
σ Acceptable process Shift σ Acceptable σ Shift
1 30.2% 697,672 1 68.3% 317,311
2 69.1% 308,770 2 95.5% 45,500
3 93.3% 66,811 3 99.7% 2,700
4 99.4% 6,210 4 100.0% 63.37
5 99.98% 233 5 100.00% 0.574
6 99.9997% 3.4 6 100.0000% 0.002

SIPOC SIPOC stands for Suppliers, Inputs, Process, Output, and Customers. You obtain inputs from suppliers, add
value through your process, and provide an output that meets or exceeds your customer's requirements.
Process Mapping Steps:
1. Name the PROCESS
2. Identify the START and STOP points
3. Identify OUTPUT of process
4. Identify the CUSTOMER(s) of the process
5. Identify the SUPPLIER(s) of the process
6. Identify the INPUTS of the process
Indicate the 5 to 7 highest level steps in the process AS THEY EXIST TODAY
Six M’s Machines, Materials, Methods, Measurement, Mother Nature (environment), Man (people) – Fish bone diagram
Six Sigma Improve effectiveness and efficiency
SMART Specific, Measurable, Action oriented, Realistic, Time bound
Stage Gate
Process

Standard = square root of variance


Deviation σ &
S
STATISTICS QUANTITATIVE: asks “how many” classified into:
Types of • DISCRETE data - 3 dents per unit for example (values cannot be subdivided meaningfully)
• CONTINUOUS data – 2.007 inches in diameter for example (time, distance, cost, etc)
• Based on sampling and probability methods
QUALITATIVE;
• Deals with attributes and making inferences and conclusions.
• asks “why” i.e. cannot be measured only described.
• Not based on probability (on a scale of 1 to 10 how do you…)

SWOT Strengths Weakness Opportunities Threats (Michael Porter)


Symbols μ = Population MEAN
σ = Population STANDARD DEVIATION
σ² = Population VARIANCE
N = Total number of individual observations in the population
n = Individual observations & sample size
Xi = Individual observations in the population or sample
X = Sample MEAN
r = Correlation coefficient (scatter diagram / linear regression)
S = Sample STANDARD DEVIATION
S² = Sample VARIANCE
α = Alpha (type I error - rejecting a null hypothesis when it is actually true)
β = Beta (type II error - failing to reject a null hypothesis when the alternative hypothesis is true)
χ =Chi Square
Takt time Takt time can be defined as the maximum time allowed to produce a product in order to meet demand.
Example: customer buys 16 units a day so your takt time is 1/2 hour (1 shift), time needed to satisfy customer
demand. The customer should PULL products through the process, not you producing products in hopes that
they will sell (pushing).
TDPU Total Defects Per Unit – used for multi-step process TDPU = Inverse of RTY (Rolled Throughput Yield)
Example: Yield rates are as follows Y1=99.8% Y2=97.4% Y3=96.4% so RTY=(.998)(.974)(.964)=0.93706 - TDPU = Inverse (RTY) =
In(.93706) = 0.065
Team Roles Most important: Member, Leader, Facilitator, Recorder,& Timekeeper, Process owner,
TOC Theory Of Constraints
• Increase THROUGHPUT
• Reduce INVENTORY
• Reduce OPERATING EXPENSES
• Balance FLOW through the plant (process) Note: NOT balance capacity
The key steps in implementing an effective process of ongoing improvement according to TOC are:
0. (Step Zero) Make the GOAL explicit. Frequently, this is something like, "Make money now and in the future."
1. Identify the CONSTRAINT (the thing that prevents the organization from obtaining more of the goal)
2. Exploit the CONSTRAINTt (make sure the constraint is doing things that the constraint uniquely does, and not doing things that it
should not do)
3. Subordinate all other processes to above decision or CONSTRAINT (Let the resources who have excess capacity spend some of that
capacity to help the Constraint)
4. Elevate the constraint (if required, permanently increase capacity of the constraint; "buy more")
5. If, as a result of these steps, the constraint has moved, return to Step 1. Don't let inertia become the constraint.
• Drum - The constraint(s), linked to market demand, is the drumbeat for the entire plant.
• Buffer - Time/inventory that ensures that the constraint(s) is protected from disturbances occurring in the system.
• Rope - Material release is "tied" to the rate of the constraint(s).
TPM Total Productive Maintenance – complete elimination of
Overall Equipment = Availability x Performance x Rate of
failures, defect, waist & loss due to equipment related Effectiveness Efficency Quality Products
operations. Availability = Operation Time = Loading Time - Downtime
Loading Time Loading Time

Variance σ² & =standard deviation² (squared)



Variation Two types
• Common Cause – no undue influence
• Special Cause – one or more of the 5M’s & 1P have an undue influence
Waste, in 1. Overproduction: One of the important concepts of lean manufacturing is the pull scheduling. This
manufacturing means the production is pulled by the process above them.
Seven (7) (or 8) 2. Waiting: Waiting for processing.
3. Higher Transportation: Transportation adds up to the lead-time, uses space and money, create delays.
4. Inappropriate Tooling (or process): The best tools (or processes) for the particular purpose are
required, money is not saved buying cheap tooling or using a inefficient process.
5. Higher WIP: Work in progress is a direct result of over production and waiting. It hides the quality
defects and other problems in the system.
6. Movement & Ergonomic Imperfections: High amount of motion in work is inefficient and affects the
health of the worker. This is a long term cost for the company.
7. Defects: No need to explain it more.
8. Underutilization Of Human Talents: By using human resources effectively every other waste (#1-7)
can be eliminated from the system. ( This is added to the 7 wastes)

You might also like