Read shainin%20amelior.pdf text version

Shainin: A concept for problem solving

Lecture at the Shainin conference Amelior 11 December 2009

Willy Vandenbrande

qsconsult www.qsconsult.be

1

Dorian Shainin (1914 ­ 2000)

· Aeronautical engineer (MIT ­ 1936) · Design Engineer for United Aircraft Corporations · Mentored by his friend Joseph M. Juran · Reliability consultant for Grumman Aerospace (Lunar Excursion Module) · Reliability consultant for Pratt&Whitney (RL-10 rocket engine) · Developed over 20 statistical engineering techniques for problem solving and reliability · Started Shainin Consultants in 1984, his son Peter is current CEO.

2

qsconsult

Dorian Shainin and ASQ

· 15th ASQ Honorary Member (1996) · First person to win all four major ASQ medals · In 2004 ASQ created the Dorian Shainin Medal

­ For outstanding use of unique or creative applications of statistical techniques in the solving of problems related to the quality of a product or service.

3

qsconsult

Dorian Shainin

· Not very well known outside USA (compared to Deming, Juran) · 1991: Publication of first edition of "World Class Quality" by Keki Bothe · 2000: Second edition (Keki and Adi Bothe) · Books brought attention to Shainin methods, but are very biased.

4

qsconsult

Problem Solving

· Focus is on variation reduction

LSL USL After

Before

LSL = Lower Specification Limit USL = Upper Specification Limit

5

qsconsult

Problem Solving

· But also ...

LSL After Before

6

qsconsult

Basic Shainin assumption

· The pareto principle of vital few and trivial many. · Only a few input variables are responsible for a large part of the output behavior.

­ Red XTM ­ Pink XTM ­ Pale Pink XTM

· Problem solving becomes the hunt for the Red XTM

7

qsconsult

Shainin tools

· Recipe like methods / statistics in the background · Comparing extremes allows easier detection of causes ­ BOB ­ WOW

Best of Best Worst of Worse

· Non parametrics with ranking tests in stead of calculations with hypothesis tests · Graphical Methods · Working with small sample sizes · The truth is in the parts, not in the drawing: let the parts talk!

8

qsconsult

Preliminary activities

· Define the critical output variable(s) to be improved (called problem Green Y®) · Determine the quality of the Measurement System used to evaluate the Green Y®

­ A bad measurement system can in itself be responsible for excessive variation ­ Improvements can only be seen if they can be measured

9

qsconsult

20 ­ 1000 variables

Clue generating

Components Search

Multi-Vari chart

Paired Comparisons

Product / Process Search

5 ­ 20 variables

Variables Search

Full Factorials

4 or less variables

Formal Doe tools Validation

No interactions

B vs C

Interactions

Optimization

Scatter Plots

RSM methods

Assurance Control Ongoing control

Positrol

Process Certification

Precontrol

Overview of Shainin tools

qsconsult

10

Source: World Class Quality ­ 2nd edition

General comments

· Gradually narrowing down the search · Clear logic

­ Analyzing ­ Improving ­ Controlling

· Not all tools are "Shainin" tools · "What's in a name?"

­ Positrol versus Control Plan ­ Process Certification versus Process Audit

11

qsconsult

Tool details

· Overview of methods · More info on B vs CTM and Scatter Plots in workshops · Some more detail on

­ Multi-Vari chart ­ Paired ComparisonTM and Product/Process Search ­ Pre Control

12

qsconsult

Clue Generating / Multi-Vari Chart

Objective Application Principles Sample Size Comments

Understand the pattern of variation Define areas where not to look for problems Allow a more specific brainstorm Problem type: excess variation Wide applicability Divide total variation in categories Search for causes of variation in the biggest category first Samples taken in production on current process Could be a big measurement investment Very useful tool and best applied before brainstorming causes on excess variation

13

qsconsult

Multi-Vari Chart

· Breakdown of variation in 3 families:

­ Positional (within piece, between cavities, ...) ­ Cyclical (consecutive units, batch-to-batch, lot-tolot) ­ Temporal (hour-to-hour, shift-to-shift, ...)

14

qsconsult

Multi-vari Chart

· If one family of variation contains a large part of total variation, we can concentrate on investigating variables related to this family of variation.

15

qsconsult

Clue Generating / Component SearchTM

Objective Application Principles Sample Size Comments

Disassembly / reassembly requirement limits application. Find the component(s) of an assembly that is (are) responsible for bad behavior Problem type: assembly does not perform to spec Limitation: Disassembly / Reassembly must be possible without product change Select BOB and WOW unit Exchange components and observe behavior. Components that change behavior are Red X comp 2 = 1 BOB and 1 WOW

16

qsconsult

Clue Generating / Paired ComparisonTM

Find directions for further investigation

Objective Application Principles Sample Size

Practical application of "let the parts talk" Problem type: occasional problems in production flow Select pairs of BOB and WOW units Look for differences Consistent differences to be investigated further 5 to 6 pairs of 1 BOB and 1 WOW

Comments

17

qsconsult

Paired ComparisonsTM: method

· Step 1: take 1 good and 1 bad unit

­ As close as possible in time ­ Aim for BOB and WOW units

· Step 2: note the differences between these units (visual, dimensional, mechanical, chemical, ...). Let the parts talk! · Step 3: take a second pair of good and bad units. Repeat step 2

18

qsconsult

Paired ComparisonsTM: method

· Step 4: repeat this process with third, fourth, fith, ... pair until a pattern of differences becomes apparent. · Step 5: don't take inconsistent differences into account. Generally after the fith or sixth pair the consistent differences that cause the variation become clear.

19

qsconsult

Clue Generating / Product/Process Search

Objective Application Principles Sample Size Comments

Tukey test is alternative for t-test Widely applicable method Problem: available data (process parameters)

20

qsconsult

Preselection of variables out of a large group of potential variables Problem type: Various types of problems

Select sets of BOB and WOW units ­ batches - .. Add product data / process parameters and rank Apply Tukey test to determine important parameters 8 BOB and 8 WOW units / batches

Product/Process Search: example

· Transmission assemblies rejected for noise. · Components search shows idler shaft as responsible component · One of the parameters of idler shaft is "out of round" · 8 good / 8 bad units selected and measured for "out of round"

21

qsconsult

Product/Process search: example

Out of round good units (mm) 0.015 0.018 0.014 0.022 0.017 0.019 0.011 0.007 Out of round bad units (mm) 0.019 0.018 0.016 0.023 0.024 0.023 0.021 0.017

22

qsconsult

Tukey test procedure

· Rank individual units by parameter and indicate Good / Bad. · Count number of "all good" or "all bad" from one side and vice versa from other side. · Make sum of both counts. · Determine confidence level to evaluate significance.

23

qsconsult

Tukey test confidence levels

Total end count 6 7 10 13 Confidence 90% 95% 99% 99.9%

24

qsconsult

Tukey test: example

Good 0.007 0.011 0.014 0.015 0.016 0.017 0.018 0.019 0.021 Bad

Top end count (all good)

4

0.017 0.018 0.019 0.022

Overlap region

0.023 0.023 0.024

Bottom end count (all bad)

3

25

qsconsult

Tukey test: example

· Total end count = 4 + 3 = 7 · 95 % confidence that out-of-round idler shaft is important in explaining the difference in noise levels.

26

qsconsult

Formal Doe tools / Variables Search

Objective Application

Determine Red XTM, Pink XTM including quantification of their effect Problem type: Various types of problems After clue generating more then 4 potential variables left

List variables in order of criticality (process knowledge) and indicate good / bad level. Swap factor settings and observe behavior. Factors that change behavior (and interactions) are red XTM, Pink XTM

Principles

Sample Size Comments

Number of tests is determined by number of variables and quality of ordering. Alternative to fractional factorials on two levels Method comparable to components search

27

qsconsult

Formal Doe tools / Full Factorials

Objective Application Principles Sample Size Comments

Determine Red XTM, Pink XTM including quantification of their effect Problem type: Various types of problems After clue generating 4 or less variables left Classical DOE with Full Factorials at two levels Main Effects and interactions are calculated Number of tests is determined by number of variables k (2k test combinations) Well established method

28

qsconsult

Formal Doe tools / B(etter) vs C(urrent)TM

Objective Application Principles Sample Size Comments

Create new process using optimum settings and compare optimum with current.

3 B and 3 C tests (each test can involve several units ­ test of variation reduction) All 3B's must be better than all 3C's Quick validation that works well with big improvements Validation of Red XTM, Pink XTM Problem type: Various types of problems

29

qsconsult

Optimization / Scatter Plots

Objective Application Principles Sample Size Comments

Graphical method that could easily be transformed to a statistical method Fine tune best level and realistic tolerance for Red XTM, Pink XTM if no interactions are present Problem type: Variation Reduction and optimizing signal

Do tests around optimum and use graphical regression to set tolerance

30 tests for each critical variable

30

qsconsult

Optimization / Response Surface Methods

Objective Application Principles Sample Size

Method developed by George Box Fine tune best level and realistic tolerance for Red XTM, Pink XTM if interactions are present Problem type: Variation Reduction and optimizing signal

Evolutionary Operation (EVOP) to scan response surface in direction of steepest ascent

Depends on variables and surface.

Comments

31

qsconsult

EVOP example

32

qsconsult

Control / Positrol

Objective Application Principles Sample Size

Can be compared with a Control Plan Assuring that optimum settings are kept Problem type: all types

Table of What, How, Who, Where and When control has to be exercised.

Checking frequency in the When column

Comments

33

qsconsult

Control / Process Certification

Objective Application Principles Sample Size Comments

Mix of 5S, Poka-Yoke, instructions, ISO 9000, audits,... Make overview of things that could influence the process and install inspections, audits, ... Checking frequency to be determined Eliminating peripheral causes of poor quality Problem type: all types

34

qsconsult

Control / Pre Control

Objective Application Principles Sample Size Comments

Alternative to classical SPC Traffic lights system Very practical method

35

qsconsult

Continuous checking of the quality of the process output Problem type: control variation and setting of the process Divide total tolerance in colored zones and use prescribed sampling and rules to control the process. Checking frequency to be determined

Pre-Control: chart construction

1/4 TOL

1/4 TOL

USL

½ TOL TARGET

LSL

36

qsconsult

Pre-control: use of chart

1. Start process: five consecutive units in green needed as validation of set-up. 2. If not possible: improve process. 3. In production: 2 consecutive units 4. Frequency: time interval between two stoppages (see action rules) / 6.

37

qsconsult

Pre-control: action rules

Result of samples 2 units in green zone 1 unit in green and 1 unit in yellow zone 2 units in same yellow zone 2 units in different yellow zone 1 unit in red zone Action Continue Continue Correct Stop and act Stop and act

After an intervention: 5 consecutive units in green zone

38

qsconsult

Pre-control: example

Start

qsconsult

Correct

Start

Time

39

QS Consult

Willy Vandenbrande, Master TQM ASQ Fellow - Six Sigma Black Belt Montpellier 34 B - 8310 Brugge België - Belgium Tel + 32 (0)479 36 03 75

E-mail [email protected] Website www.qsconsult.be Willy Vandenbrande

qsconsult www.qsconsult.be

40

Information

40 pages

Report File (DMCA)

Our content is added by our users. We aim to remove reported files within 1 working day. Please use this link to notify us:

Report this file as copyright or inappropriate

1892


Notice: fwrite(): send of 203 bytes failed with errno=104 Connection reset by peer in /home/readbag.com/web/sphinxapi.php on line 531