Read Slide 1 text version

QBASE

Engineering

Shainin's Methods Practical Design of Experiment

Shainin's DOE Tools I List of Clue-Generating Tools

a. Multi-Vari Analysis b. Concentration Charts c. Component Search d. Paired Comparison e. Product/Process Search

II. Formal Design of Experiment

Technique to Characterize a Product / Process

a. Variable Search b. The Full Factorial c. B versus C

III. DOE Optimization a. Scatter Plot - to Achieve Realistic Specification and Tolerances b. Response Surface Method (RSM) - to Optimize Interactions V. Transition from DOE to a. Positrol: Holding the gains B. Process Certification: Eliminating Peripheral Causes of Poor Quality

Statistical Process Control

Statistical Process Control:

a. Pre-control Simple & effective technique of Process Control

For detail contact: [email protected] Or [email protected]

© QBASE Engineering Sdn FAQ's and Quality Tips page 1

QBASE

Engineering

I List of Clue-Generating Tools a. Multi-Vari Analysis

b. Concentration Charts c. Component Search d. Paired Comparison e. Product/Process Search

CLUE GENERATION TOOLS Multi-vary Experiments · to reduce a larger number of unknowns and unmanageable causes of variation into a much smaller of related variables containing the Red X (i.e most dominant cause.) · It is a graphical technique to zoom in to the most likely cause of the problem by eliminating non-contributing causes of variation. · In most application, multi-vari technique acts as the first filter which later followed by other clue generation tools. Concentration Charts · Sequel to Multi-vary Experiments. It is used to pinpoint repetitive defects by location or components · Determines how a product/process is running; a quick snapshots without massive historical data and can be substitute for replace process capability studies in some white collar applications · Normally used when the Red X is " within-unit" · Min 9 to 15 or until 80% of historic variation is captured. Component Search · From hundred of thousands of components/ subassemblies, home in the Red X, capturing the magnitude of ALL important main effects and interaction effects. · Normally used when there are two differently performing assemblies ( labeled as "good" and "bad") with interchangeable components (electric motors, suspension system of a car..)

· Typically use at prototype, engineering pilot run, production pilot run, or in field.

· Require only 2 samples ­ One(1) "good" unit and one(1) "bad" unit

For detail contact: [email protected] Or [email protected]

© QBASE Engineering Sdn FAQ's and Quality Tips page 2

QBASE

Engineering

CLUE GENERATING TOOLS Paired Comparison · Used to identify the Red X when the good and bad units, assembly or subassembly cannot disassemble and reassemble without damaging or destroying or radically changing the good and bad units properties. · Use in situation where there are two differently performing assemblies ( labeled as "good" and "bad") incapable of interchangeable of the components. · Commonly used in New product and/or process design production, field, support services, administrative works, farms, hospital, and schools. · It is a logical sequence to component search, when the Red X, distilled from the system, subsystem, and subassembly component search, cannot be disassemble any further. · Sample size required : 12 to 16 - 6 to 8 "good" units and 6 to 8 "bad" units in rank order. Product Process Search ·To identify important product variables identified with paired comparison. · To identify important process variables associated with 8 good and 8 bad products. ·Commonly used in situation where it is difficult to isolate important process variables with multi-vari ·Typically used during prototype, engineering pilot run, production pilot run, in field or in full production ·Sample size required: Sufficient units through a process to produce 8 "good" units and 8 "bad" units and their associated process parameters

For detail contact: [email protected] Or [email protected]

© QBASE Engineering Sdn FAQ's and Quality Tips page 3

QBASE

Engineering

II. Formal Design of Experiment Technique to Characterize a Product / Process a. Variable Search b. The Full Factorial c. B versus C

FORMAL DESIGN OF EXPERIMENT TECHNIQUES Variable Search ·Excellent problem prevention tools normally used to Pinpoint the Red X, Pink X etc. · Capture the Magnitude of all important main effects and interaction effects. Of the red X, pink X etc. · To identify any unimportant factors so that their tolerances can be liberated to reduce cost. · Normally used when there are High Number of variable to investigate ( 5 to 20 variables). · Application in white collar work (off-line quality control). ·Typically used in R& D , Development engineering, Product Process Characterization in Production . · For pinpointing the Red X after Multi-Vari or Paired Comparison experiments have been conducted. ·Sample Size required - 1 to 20

PUREST

·Full Factorial ·Variable Search ·Latin Square ·Plackett-Burman ·Fractional Factorial ·Taguchi Orthogonal Array

Most CONTAMINATED

For detail contact: [email protected] Or [email protected]

© QBASE Engineering Sdn FAQ's and Quality Tips page 4

QBASE

Engineering

FORMAL DESIGN OF EXPERIMENT TECHNIQUES B versus C · Basically used as Verification Tool. · To predict how much better a given product or process is than another, with confidence of 90% or higher. · To assure the permanency of an improved product or process over a previous one. · To select one product or process over another, even if there is not improvement in quality, because of some tangible benefits, such as cost or cycle time. ·To evaluate more than just two product, processes, materials (B,C,D,E etc) simultaneously

Full Factorial

·To determined which of the 2,3 or 4 variables - filtered through one or more clue-generation techniques- are important and which are unimportant; ·To open up tolerance of the unimportant variables to reduce costs; ·To quantify the magnitude and desired direction of the unimportant variables and their interaction effects, and to tighten the tolerance of these variables to achieve a Cp, Cpk = 2.00 and more; ·Investigative tool at design or prototype stage where samples are limited for other clue-generating tools.; ·Note: Even though Full factorial experiment is a problem-solving tool, it is not recommended to use it as a start of a problem investigation bypassing other clue generation tools.

For detail contact: [email protected] Or [email protected]

© QBASE Engineering Sdn FAQ's and Quality Tips page 5

QBASE

Engineering

III. DOE Optimization a. Scatter Plot - to Achieve Realistic Specification and Tolerances b. Response Surface Method (RSM) - to Optimize Interactions

DOE FOR OPTIMIZATION Scatter plot ·Used to establish Realistic Specifications and Realistic Tolerances. · Used to adjust or Tighten the Tolerances of the important product/process or Red X variables to achieve high Cpk's. · Open up the Tolerances of the unimportant variables to reduce cost. Response Surface Methods (RSM) ·To determine the BEST combinations of levels of two or more INTERACTING input variables ( identified in previous DOE experiments) to achieve a maximum, minimum , or optimum Green Y ( Response, output and Green Y are the same terms).

For detail contact: [email protected] Or [email protected]

© QBASE Engineering Sdn FAQ's and Quality Tips page 6

QBASE

Engineering

V. Transition from DOE to a. Positrol: Holding the gains

Control:

Vi) Statistical Process a. Pre-control Simple & effective technique of Process Control

Statistical Process Control

B. Process Certification: Eliminating Peripheral Causes of Poor Quality

TRANSITION FROM DOE TO SPC Positrol The POSITROL plan determines: a) WHAT the variable characterized and optimized through previous DOE experiment. B) WHO should be performing the monitoring, measuring and recording each of important variables. C) HOW determines the correct instrumentation to measure these important variables( observing the 5:1 rule ). D) WHERE determine optimum location of measuring the process parameters so that it truly reflects the correct value. E)WHEN is the frequency of measurement, determine initially by engineering judgment, but later by pre-control. ·Use process certification to eliminate the Peripheral Causes of Variation and Poor Quality such as: ·Management/supervision inadequacy ·Violation of Good Manufacturing Practices (GMP) ·Plant/Equipment inattention ·Environment Neglect ·Human Shortcomings STATISTICAL PROCESS CONTROL SPC: Pre-Control · The use of simple and cost effective pro-control chart and reaction plan to ensure the process sustain the improvement achieved. ·Typically at the last stage of improvement process.

Process Certification

For detail contact: [email protected] Or [email protected]

© QBASE Engineering Sdn FAQ's and Quality Tips page 7

QBASE

Engineering

Problem Solving Framework - linking all the Shainin's Tools

Objectives ·Proper Understanding and defining the problem at hand. ·To pinpoint the problem ·Improve the resolution of the problem ·Understanding historical background of the problem ·To identify all possible causes of the problem and sources of variation ·To identify the possible variables/factors related to the problem ·To identify the possible process variables/factors related to the problem ·To identify the possible parts/components related to the problem ·To validate the possible parts/components related to the problem ·To specified the optimize the Red X ( significant cause(s) ) with proper tolerances. ·To maintain the improvement achieved through well defined series of control mechanisms ·Manage the improved / validate process ·Daily management of the process ·Measure scatter plot ·Use Likert Scale to convert attributes into variables ·Trends (Pareto, Defect rate, Cost ) ·Multi-vari ( including concentration chart) · Component Search ·Paired Comparison ·Product/Process Search ·Variable Search · Full Factorial · B vs.C Common Shainin's DOE Tools

Problem Solving Steps 1.Define the problem ( Green Y) 2. Quantify and measure the Green Y 3.Problem History (problem history, defect rate, cost) 4.Generate Clues

5.Formal Design of Experiment (DOE)

6. Turn Problem on and Off ­ ensuring permanence of improvement 7. Establish realistic specification and Tolerances (optimize) 8. 8. "Hold" the process improvement gains 9. Hold the Gain with SPC

> B vs. C

· Scatter Plot · Response Surfaced Method (RSM) ·Positirol

Pre-control

For detail contact: [email protected] Or [email protected]

© QBASE Engineering Sdn FAQ's and Quality Tips page 8

QBASE

Engineering

Relationship between Green Y and Red X : Pareto Principle

Green Y

50% The Vital Few The Trivial Many

1

2

3

4 5 6 Causes/variable/factors/component/parts

7

Red X

Pink X

Pale Pink X

Note: Solving for the Red X, Pink X and Pale Pink X can: 1. Reduce variation 3. Eliminate the Green Y (problem) 2. Achieve Cpk of 2.00 to 10.00 with one, two or three experiments

For detail contact: [email protected] Or [email protected]

© QBASE Engineering Sdn FAQ's and Quality Tips page 9

QBASE

Engineering

Shainin's Methods Problem Solving Steps

Searching For Red X: Problem Solving Steps

Problem Green Y Measurement System /Discrimination Ratio Accuracy, Bias ,precision 5:1 Accuracy Variation Family Known? 1 Multi-Vari Experiment

Scatter Plot/Multivari ·Within Instruments ·Instrument-to-instruments ·Operator-to-operator ·Likert Chart

Unit-to-Unit variation

N

Within Unit variation

N

Time-to-Time variation

Y

2

Component search Experiment Assemble/ Reassemble

Y

Y

Next page

Y

Next page

Progressive Disassembly 4

N

Green Y Constant?

Y

3

Part/Component Related Paired Comparison 5

Variable Search Full Factorial Experiments

N

Red X Identified?

Y

4

B vs.C Experiment

5

Response Surface Method Experiment

Y

Interaction present?

N

Scatter-Plot Experiment

6

Positrol

Process Certification 7

Pre-control

For detail contact: [email protected] Or [email protected]

FAQ's and Quality Tips page 10

© QBASE Engineering Sdn

QBASE

Engineering

Shainin's Methods Problem Solving Steps

Searching For Red X: Problem Solving Steps

From Previous Page

Unit-to-Unit variation

N

Within Unit variation

N

Time-to-Time variation

y

Concentration Chart

Refer to Previous Page Next page

Paired Comparison

Variable Search Full Factorial Experiments

N

Red X Identified?

Y

B vs.C Experiment

Response Surface Method Experiment

Y

Interaction present?

N

ScatterPlot Experiment

Positrol

Process Certification

Pre-control

For detail contact: [email protected] Or [email protected]

© QBASE Engineering Sdn FAQ's and Quality Tips page 11

QBASE

Engineering

Shainin's Methods Problem Solving Steps

Searching For Red X: Problem Solving Steps

From Previous Page

Unit-to-Unit variation

Within Unit variation

Time-to-Time variation

Refer to Previous Page

Refer to Previous Page

Product Process Search

Variable Search Full Factorial Experiments

N

Red X Identified?

Y

B vs.C Experiment

Response Surface Method Experiment

Y

Interaction present?

Y

ScatterPlot Experiment

Positrol

Process Certification

Pre-control

For detail contact: [email protected] Or [email protected]

© QBASE Engineering Sdn

FAQ's and Quality Tips

page 12

Information

Slide 1

12 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

2142


Notice: fwrite(): send of 205 bytes failed with errno=104 Connection reset by peer in /home/readbag.com/web/sphinxapi.php on line 531