SixSigma GB+Module

download SixSigma GB+Module

of 50

Transcript of SixSigma GB+Module

  • 7/29/2019 SixSigma GB+Module

    1/50

    6 SigmaQUALITY The Need

    With the advent of Globalization every organization relentlessly

    pushes themselves for the processes that gives them error freeproducts in a competitive environment. With the growth of

    competitive environment - error free products acts as aQualifying parameter only. However the survival demands more.

    What is more?

    More is defined by the Customer and the competitive environment.

    - Critical to Customer- Critical to Delivery

    - Critical to Cost

    - Critical to Process etc.

  • 7/29/2019 SixSigma GB+Module

    2/50

    6 SigmaQUALITY Early Growth

    Motorola suffered a severe downturn when a Japanese manufacturer invaded

    the chip market in 1985-86. In 1987, Motorola started process improvementprojects to counter the invasion. Initial finding was the process cycle time,

    but quality fell off the radar. Later the managers began to recognized the

    quality problem was due to wait times, inventory mismanagement etc.

    Motorolas biggest customer Ford Motors took considerable interest in theprocess improvement and Six Sigma kick-off.

    Lots of people participated in the invention of Six sigma over a long period

    of time. The Six Sigma methodology package was rolled out in 1987-88.

    Motorola was honored with Malcolm Baldrige National Quality Award in 1988.

  • 7/29/2019 SixSigma GB+Module

    3/50

    6 SigmaQUALITY Project Priority

    Priority #3

    Cause Known

    Solution Unknown

    Priority #1

    Cause Unknown

    Solution Unknown

    Cause Known

    Solution Known

    Priority #2

    Cause Unknown

    Solution Known

    Cause

    Soluti

    on

    Known Unknown

    Unknown

    K

    nown

  • 7/29/2019 SixSigma GB+Module

    4/50

    6 SigmaQUALITY Methodology

    Definewhats important

    Measure

    how were doing

    Analyzewhats wrong

    Improveby fixing whats wrong

    Controlto guarantee performance

  • 7/29/2019 SixSigma GB+Module

    5/50

    AnalyzeOpportunity

    PlanImprovement

    FocusImprovement

    DeliverPerformance

    ImprovePerformance

    6 SigmaQUALITY Parallel-Lean Six Sigma

    Lean Six Sigma

    Lean Six Sigma is combination of Lean Method and Six Sigma. The method

    builds on the knowledge, methods and tools derived from decadesof operational improvement research and implementation.

    Lean approaches focus on reducing cost through process optimizations.

    Its aim is effectiveness, not just efficiency. Broadly Lean Six Sigma is for

    operational improvements - refining existing processes to reduce cost.

    Six Sigma is about meeting customer requirements, stakeholder expectation

    and improving quality by measuring and eliminating defects.

  • 7/29/2019 SixSigma GB+Module

    6/50

    6 SigmaQUALITY Six Sigma Jargon

    Champion -The Champion is a person responsible for instilling the vision of Six Sigma and communicating

    it across the firm. They are usually the upper management or executive officers. The assistin dedicating the resources and choosing the projects.

    Master Black Belt -The MBB people are those, who have extensive experience in Six Sigma methodology.

    MBBs acts as a coach to its team member in project planning and result evaluation.

    The BB people leads the project on fulltime basis. BBs are certified people with hands on

    Six Sigma projects. They have strong understanding of statistical methods of data collection

    and analysis. They are the project managers and are responsible for all traditional roles.

    Black Belt -

    Green Belt -The GBs acts as an assistant to BBs in their job. GBs have the basic understandings of

    statistic but dont have the expertise and experience like BBs. GBs does the legwork of BBsin project realization. However, they lead the project on a part time basis.

  • 7/29/2019 SixSigma GB+Module

    7/50

    6 SigmaQUALITY Define Phase

    Involves:

    ProjectCharter

    To develop a Project Charter. Objective Develop focus and purpose of the team.

    VOC

    Use Voice of Customer to identify customer needs.

    Objective Identify the actual pain / opportunity.

    CTQs

    Translate customer needs into Critical To Quality elements. Objective Identify Project CTQs.

    ProcessMap

    Translate customer needs into Critical To Quality elements. Objective Identify Project CTQs.

  • 7/29/2019 SixSigma GB+Module

    8/50

    6 SigmaQUALITY Define Phase

    Project Charter: Elements

    BusinessCase

    Problem

    Statement

    GoalStatement

    Scope

    Milestone

    - Why is the project worth doing

    - Consequence of not doing the project

    - Description of problem

    - What is not meeting customers requirement

    - Improvement the team desires

    - What lies in scope and whats out of scope

    - Project plan with timeline

  • 7/29/2019 SixSigma GB+Module

    9/50

    6 SigmaQUALITY Define PhaseProject Charter: RACI Model

    Responsibility Charting is a technique for identifying functional areas where there are

    process ambiguities . It brings the differences out in open and resolving them through a

    cross-functional collaborative effort. It defines the participation roles to different departments

    linked with the project.

    Responsible (R) doerThe doer(s) is responsible for action/implementation. The doers are those who actually

    completes the task.

    Accountable (A) the buck stops hereThe accountable persons are those who are actually answerable for activity.

    This includes yes or no authority. Only one A is assigned to an action.

    Consult (C) in the loopThe consultants are consulted prior to a final decision or action.

    Inform (I) keep in the pictureThe communication to this individual is a one-way. They are just informed about the changes

    and are expected to take action as a result of the changes.

  • 7/29/2019 SixSigma GB+Module

    10/50

    6 SigmaQUALITY Define PhaseRACI Charting: Example

    RACI Closing Guideline:

    1. Place the Accountability & Responsibility at lowest feasible level.

    2. There can be only one accountable individual per activity.

    3. Authority must accompany accountability.

    4. Minimize the number of Consults & Informs.

    5. Horizontal & Vertical Analysis of RACI chart indicates flaws in roles/responsibilities.

    Mother Father Sam Jenny Clark Kids

    A C R

    I I A R

    C A/R R

    C A/R R

    C A/R

    C A R

    Feed the Dog

    Play with Dog

    Morning Walk

    Evening walk

    Clean the Dog

    Clean the Table

  • 7/29/2019 SixSigma GB+Module

    11/50

    6 SigmaQUALITY Define PhaseVOC / CTQs if we can find it, we can focus on it

    Voice of customer (internal/external) is tracked through various means. Customer Surveys

    and Complaints are generally used for external customers while executive level discussions

    are intended for refining/improving internal processes. Benchmark data is essential for

    process developments/innovations with respect to peers.

    Customer Surveys Complaints Executive leveldiscussions

    Benchmark Data

    Example- CTQs

    Critical To Quality(CTQ)

    Dimensions

    Critical To Delivery(CTD)

    JIT Compliance

    Critical To Cost(CTC)

    RM/Scrap/Inventory

    Critical To Process(CTP)

    Process Cycle Time

    Definition of CTQs changes as per projects, However, it should be very clear and straight-

    forward as it forms the base of project.

  • 7/29/2019 SixSigma GB+Module

    12/50

    6 SigmaQUALITY Define PhaseProcess Mapping: y = f(x, x, x ..)

    Process Mapping begins with SIPOC map. Once SIPOC is prepared, Process part is again

    fragmented into a high level process diagram involving all work centers that affects Output.

    Majority of this process comes underMeasure Phase, but to prioritize and link CTQs to

    various process inputs.

    Supplier Input Process Output Customer

    High level Process Map

    S I P O C

  • 7/29/2019 SixSigma GB+Module

    13/50

    6 SigmaQUALITY Measure PhaseTools:

    Measure Phase is a pioneer stage in quantifying, qualifying and validating the Six Sigma

    needs. It is in this phase that a more refined data evolves in terms of CTQs. And the phase

    translates the physical behavior into statistical problem. The phase also makes an

    assessment of current process performance.

    Many companies have their own methodology of recording day-to-day process data

    however, methods of measurement revolves around above stated five tools.

    Process Mapping YX MatrixMeasurement

    Systems Analysis(MSA)

    Capability Analysis

  • 7/29/2019 SixSigma GB+Module

    14/50

    6 SigmaQUALITY Measure Phase

    High Level Process Mapping: Example-Drilling

    A high level process mapping is very important in understanding the interaction of various

    process interaction and linkage of CTQs with the process centers. Majority of concentration

    is put on in eliminating the excess or the wastage of cycle time as against to the optimalrequirement.

    Start Work OrderQTY/PartRefer theDrawing

    Use Jig

    Jig

    Required?

    Go forMarking

    Prepare

    Tooling

    Start

    Drilling

    Is

    Drilling

    OK

    End

    Rework

    Possible

    Rejection

    Report

    Rework

    Report

    Rework Area

    Rejection

    No

    YesNo

    Yes

    No

    Yes

  • 7/29/2019 SixSigma GB+Module

    15/50

    6 SigmaQUALITY Measure PhaseYX Matrix:

    The YX Matrix is an exhaustive process study to zero in on to the process characteristic

    that has greater impacts on CTQs. Hence, we deduce the weight of different process

    impacting CTQs.

    Output Variable (Y)

    # Rank=

    Score x Weight

    Weld

    Strength

    Weld

    Appearance

    Pinhole

    Density

    Weld

    FlexibilityRank

    Weight 10 8 7 4

    Weld

    Frequency8 4 7 5 181

    Power

    Amplitude8 8 4 9 208

    Initial Gap 5 2 3 4 103

    Contact

    Pressure2 7 10 2 154I

    nputVariable(X)

  • 7/29/2019 SixSigma GB+Module

    16/50

    Process

    Variation

    Actual ProcessVariation

    MeasurementProcess Variation

    Variation due toOperator Variation due toEnvironmentVariation due to

    Instrument

    6 SigmaQUALITY Measure PhaseMeasurement System Analysis:

    MSA is used to separate the variations of measurement system in the process.

    Process MeasurementProcessInput Output

    Repeatability: Variation when one person repeatedly measures same unit with same system

    Reproducibility: Variation when two people measuring same unit with same system

    Tolerance: Expectation of error range

  • 7/29/2019 SixSigma GB+Module

    17/50

    6 SigmaQUALITY Measure PhaseGage R & R Analysis: Example

    Gage R & R analysis presented above seems to be complicated , but in fact its not. It is simply aprocess where Repeatability & Reproducibility is measured.

    R & R < 10% :

    Satisfactory

    R & R 10 - 30% :

    May be Satisfactory

    R & R > 30% :

    Unsatisfactory

  • 7/29/2019 SixSigma GB+Module

    18/50

    6 SigmaQUALITY Measure PhaseCapability Analysis:

    Process

    Incapable

    Process CapabilityCp

    Cp is the ratio of tolerance width to short termprocess spread.

    Estimates instantaneous capability of theprocess.

    Process PerformanceCpk

    Cpk is the ratio of the distance measuredbetween process mean and specificationlimits closer to half the total process spread.

    Estimates measure of capability of a processto meet established customer requirement.

    LSL USL

    Cp

  • 7/29/2019 SixSigma GB+Module

    19/50

    6 SigmaQUALITY Measure PhaseCp & Cpk :

    Cpk = Cp : Process Mean is on target.

    Cpk = 0 : Process Mean falls on one of the specification limits. 50% of process falls

    beyond specified limits.

    Cpk < -1 : Process Mean is completely out of specified limits.

    Cp = 2 : It implies the accomplishment of short term objective.

    Short Term process capability = 3 x Cp.

    At Cp = 2, Process Capability = 6

    Cp =USL LSL

    6STCpl =

    X LSL3ST

    -Cpu =

    USL X3ST

    Cp Cpk

    -

    Cpk = min {Cpl, Cpu}

    6 Si

  • 7/29/2019 SixSigma GB+Module

    20/50

    6 SigmaQUALITY Analyze PhaseObjective: Brainstorming

    Determine X that causes costly defects. Identify the various Hypothesis.

    Select tools to approve/disprove the Hypothesis.

    Perform statistical test and check for significance.

    If cause is statistically and practically significant recommend for Improve phase.

    Analyze phase is one of the most important phase of any project. Process data obtained inMeasure phase is analyzed and cross functional teams are consulted for brainstorming

    sessions. The session particularly aims in capturing ideas of CFTs that will help in narrowing

    the scope of issue process flaws. However, when a very large CFTs are concerned then theideas are not always uniform in structure. To bring uniformity Nominal Group Process

    technique is used. NGP implies the segregation of the ideas and grouping them in accordanceto the process/department to which it belongs. This is called as Affinity Diagram of ideas.

    After the brainstorming sessions, we arrive at number of causes. These causes are put

    together further analysis in the Improve phase.

    6 Si

  • 7/29/2019 SixSigma GB+Module

    21/50

    6 SigmaQUALITY Analyze PhaseFishbone Diagram:

    Fishbone diagram is a Cause & Effect (C&E) diagram which resembles the shape of a fish

    bone. It is also called as Ishikawa diagram, named after its creator Kaoru Ishikawa .

    6 Si

  • 7/29/2019 SixSigma GB+Module

    22/50

    6 SigmaQUALITY Analyze PhaseFMEA:

    The main objective of FMEA is to identify ways the product or process can fail and to plan

    in order to prevent those failure. FMEA in general supports Analysis Phase.Characteristics of FMEA:

    1. Identify potential failure modes.

    2. Rate the severity of their effect.

    3. Identify potential causes.

    4. Evaluate objectively the probability of occurrence of these causes.5. Evaluate the ability to detect the causes.

    6. Rank order process deficiencies.

    7. Focus on eliminating / controlling these variables.

    8. Evaluate objectively the probability of occurrence of these causes.

    Severity x Occurrence x Detection = Risk Priority Number (RPN)RPN > 80 Critical need for immediate action.

    Severity > 5 Safety related defects.High RPN & Low Detection Flaws in internal tests.

    High Occurrence Poor process capability.

    Detection (1 to 10) Good Control to Poor Control

    6 Si

  • 7/29/2019 SixSigma GB+Module

    23/50

    6 SigmaQUALITY Analyze PhaseFMEA: Example

    FMEA is a dynamic document and each process owner keeps updating it whenever required.

    Process Action Revised Result

    ProductFunction

    PotentialFailure

    Mode

    PotentialEffects

    ofFailure

    Severity

    (1-10)

    PotentialCauses

    Occurrence

    (1-10)

    CurrentControl

    Detection

    (1-10)

    RPN

    Recommended

    Action

    Responsibility&

    TargetDates

    ActionTaken

    Severity

    1-10

    Occurrence

    (1-10)

    Detection

    1-10

    RevisedRPN

    Stock

    Inventory

    Wrong

    Location

    Delayin

    Finding

    5

    Correct

    LocationFull

    7

    Checkedtwic

    e

    ayear

    9315

    MonthlyStoc

    k

    Audit

    Raghu,

    10th

    July2011

    AuditPlan

    Developed

    Stock

    Inventory

    Damaged

    Insufficient

    S

    tock

    7

    SupplierDefect

    3

    Inc

    oming

    Inspection

    8168

    Pr

    ocess

    Ins

    ection

    Padm

    avati,

    10th

    July2011

    Plan

    Dev

    eloped

    StockInventory

    Damaged

    Insufficient

    Stock

    7

    HandlingError

    5

    Standard

    Procedure

    9315

    ReviewSOP

    Sushmit,

    5th

    July2011

    SOPRevised

    6 Si

  • 7/29/2019 SixSigma GB+Module

    24/50

    6 SigmaQUALITY Analyze PhasePareto Analysis:

    Vilfredo Pareto, a 19th century Italian economist who discovered that 80% of the land in Italy

    was owned by 20% of the population, established the principle. Later it was redefined as

    80% of the problems has their roots in 20% of the causes. Hence Pareto Analysis will give

    the vital few 20% that accounts for massive 80% defects in the process.

    Example Lets consider delays in credit card processing by a bank with following reasonsand frequency of occurrences.

    Category Frequency

    No Address 9

    Illegible 22

    Current Customer 15

    No Signature 40

    Other 8

    6 Si

  • 7/29/2019 SixSigma GB+Module

    25/50

    6 SigmaQUALITY Analyze PhasePareto Analysis:

    Reorganize the frequency data in decreasing order . Then calculate the relative percentage

    on 100 percent level then calculate the cumulative percentage.

    Category FrequencyRelative

    Percentage

    Cumulative

    Percentage

    No Signature 40 43% 43%

    Illegible 22 23% 66%

    Current Customer 15 16% 82%

    No Address 9 10% 92%

    Other 8 8% 100%

    Run the Pareto analysis tool in Sigma XL or in XL select the graphical plotting tool that

    resembles the graph shown in next slide.

    6 Si

  • 7/29/2019 SixSigma GB+Module

    26/50

    6 SigmaQUALITY Analyze PhasePareto Analysis:

    Result:

    The Breaking Point: The breaking point divides the vital few from trivial many in the chart.

    Normally, the vital few starts fro 70% and above. In the above result it is No Signature,

    Illegible.

    6 Sigma

  • 7/29/2019 SixSigma GB+Module

    27/50

    6 SigmaQUALITY Analyze PhaseHypothesis Testing:

    Hypothesis testing is a value judgment made about a circumstance, a statement made about

    a population. As the name suggests, it is a technique to test the efficacy of changes in anyprocess. Lets assume that we have a data of different types of crime in a city. Now, ahypothesis can be made for the trends in a span of 10 years. But to ascertain this hypothesis

    a standard error based hypothesis testing using is conducted, which validates the hypothesis

    of crime in span of 10 years. However, in actual applications when certain data

    tends to be normal i.e. the data follows normal distributions, then standard test called ast-testis applied to validate the hypothesis. To have a better command over this test one

    needs have a clear idea of right tools and have better understanding of statistical tools.

    Common Tools:

    ANOVA

    Normal Data More thantwo group of

    equalVariance.

    Welch

    ANOVA

    Normal Data More thantwo group of

    unequalVariance

    Paired t-test

    Normal Data For matched

    two groups.

    t-test Normal Data

    Forunmatchedtwo groups

    Chi-Square

    For DiscreteData

    6 Sigma

  • 7/29/2019 SixSigma GB+Module

    28/50

    6 SigmaQUALITY Analyze PhaseHypothesis Testing: Elements

    1. Null Hypothesis (Ho): Null hypothesis is the hypothesis to be tested. In the earlierexample Ho : Crime trends in a span of 10 years.

    2. Alternate Hypothesis (H orHa):Alternate hypothesis is opposite to null hypothesis.It is assumed when null hypothesis is rejected after test. In earlier example

    H : No trends in crime in a span of 10 years.

    3. Sample t-test: Sample t-test is conducted to reject or fail to reject null hypothesis.

    4. Level of Risk : Level of risk implies the kinds of error associated while making an

    inference from hypothesis test. Two types of error can be made. The experiment

    can falsely reject a hypothesis that is true. In this case we say the error is Type Ior the (alpha) error. If test actually fails the hypothesis that is actually false

    then it is Type II or (beta) error.

    5. Decision Rule: Decision pertains to condition of rejecting or failing to reject the hypothesis.

    Predefined confidence level helps in decision.

    6 Sigma

  • 7/29/2019 SixSigma GB+Module

    29/50

    Correct Decision

    Innocent Individual GoesFree

    Incorrect Decision

    Guilty Individual Goes FreeType II or Error

    Incorrect Decision

    Innocent Individual IsDisciplined

    Type I orError

    Correct Decision

    Guilty Individual IsDisciplined

    6 SigmaQUALITY Analyze PhaseHypothesis Testing: & error

    The above example shows the condition of Type I and Type II error. The decision rule for

    above verdict is based on the confidence level of and .

    When p-value < 0.05: Reject H0 and when p-Value > 0.05 : Fail to Reject H0

    The American Trial SystemIn Truth, the Defendant is:H0: Innocent HA: Guilty

    Innocent

    Guilty

    Verdict

    6 Sigma

  • 7/29/2019 SixSigma GB+Module

    30/50

    6 SigmaQUALITY Analyze PhaseChi-Square Test: Discrete data

    Chi-Square test applies where discrete data is available for hypothesis testing. Chi-Squaretest is usually conducted in a situation where customer feedback/VOC trends are compared

    with expected trends.

    Example Lets consider a company X makes customer survey for satisfaction quarterlyand compares them with the previous expected surveys. Assume the sample size be 80.

    Category Q3-FY10 Q4-FY10

    Excellent 8 8

    Very Good 36 37

    Good 12 11

    Fair 4 7Poor 8 9

    Very Poor 12 8

    Here we will inspect the case in one variable only that is Category.H

    0

    : Result Similar and Ha

    : Results not Similar. In this test Null & Alternate hypothesisalways remains same.

    6 Sigma

  • 7/29/2019 SixSigma GB+Module

    31/50

    6 SigmaQUALITY Analyze PhaseChi-Square Test: Result

    Chi-Square test run gives following result: Chi-Sq = 1.734, DF = 5, P-Value = 0.885

    Hence, we observe that , P-Value = 0.885i.e P-Value > 0.05. Therefore, it fails to reject

    Null hypothesis ( H0 : Result Similar) and the data of Q2-FY10 & Q3-FY10 are similar.

    Which implies that company X sustained its confidence level in the market.

    6 Sigma

  • 7/29/2019 SixSigma GB+Module

    32/50

    6 SigmaQUALITY Analyze PhaseAnalysis of Variance (ANOVA):

    In 1920, Sir Ronald A. Fisher invented a statistical way to compare the data sets. Fisher

    called this method as Analysis of Variance which is popular as ANOVA. The F-ratio produced

    by ANOVA is named afterFisher. The t-test mentioned earlier had a limitation i.e. only two

    data sets can be analyzed. However, in ANOVA minimum of two data sets and maximum of

    infinite data sets can be analyzed.

    Scope of ANOVA: Whenever we want to compare more than two sets of data in order to

    conclude the better one then ANOVA plays great role.Example: Lets consider a case where a company wants to analyze the efficiency of drillingmachine. Company collects the total time taken to complete similar jobs in three locations.

    Drilling time in minute in a span of 5 days

    A B C

    15 28 26

    17 25 23

    18 24 20

    19 27 17

    24 25 21

    6 Sigma

  • 7/29/2019 SixSigma GB+Module

    33/50

    6 SigmaQUALITY Analyze PhaseAnalysis of Variance (ANOVA):

    ANOVA Test:When ANOVA test is run, we get numerical output and graphical output. Former critically

    examines the individual data for differences and later plots them for comparison.

    One-way ANOVA: A, B, C

    Source DF SS MS F P

    Factor 2 131.73 65.87 7.81 0.007Error 12 101.20 8.43

    Total 14 232.93

    S = 2.904 R-Sq = 56.55% R-Sq(adj) = 49.31%

    Numerical data examines the relation of data with each other and gives the P-value that

    indicates the confidence of test. In above case it is 0.007 and hence P-value < 0.05 which

    implies that Null hypothesis is rejected and data given are different. However, in what way

    the data differ is explained by box plot in next step.

    6 Sigma

  • 7/29/2019 SixSigma GB+Module

    34/50

    6 SigmaQUALITY Analyze PhaseAnalysis of Variance (ANOVA):

    ANOVA Test: Understanding differences

    In a box plot circle marks the average, horizontal line is median, box reaches +/- one

    standard deviation and vertical line is range of data. A close analysis of graphical plot reveals

    a lot. Location A & C are statistically similar but B differs from A & C. Location A is more

    efficient in drilling. Location B requires special attention. In one particular day at location A the

    drilling is least efficient and if rectified it will be very efficient. Location C has less consistency.

    Hence, we can conclude in a similar manner for different data.

    6 Sigma

  • 7/29/2019 SixSigma GB+Module

    35/50

    6 SigmaQUALITY Analyze PhaseRegression Analysis:

    Regression analysis is a statistical tool to analyze the relationship between quantitativevariable. It helps in predicting the process behavior when the variables are change, for

    instance how JIT is affected when inventory is reduced. However, the analysis does not

    provides the optimal level of inventory rather it gives the level of inventory required to maintain

    the desired JIT compliance. Regression transforms the physical process into a

    mathematical model that will establish the relation between dependent and independent

    variable.In above case: JIT = A (Inventory) + B , value of A and B decides how strong the relationship

    exist between JIT and Inventory. When the independent variable is more than one then it is

    called as multiple regression analysis.

    Example: Lets consider a company X is a vendor of company Y. However, company X is

    poor in managing its optimal inventory level to support JIT. But since the purchase trends ofcompany Y quarterly varies so they could not find the optimal inventory requirement.With poor inventory management the operational cost started increasing. To reduce the

    financial pain, top management wanted to analyze the effect of inventory over operational cost.

    They collected data of past one year, and wanted to know the optimal inventory level for

    particular operational cost.

    6 Sigma

  • 7/29/2019 SixSigma GB+Module

    36/50

    6 SigmaQUALITY Analyze PhaseRegression Analysis:

    Data: Operational cost against inventory level in a one year span.

    Month Operational Cost

    00000 `Inventory

    MT

    Jan 22 10

    Feb 20 8

    Mar 30 24Apr 28 20

    May 26 19

    Jun 29 23

    Jul 31 26

    Aug 37 28

    Sep 38 30

    Oct 40 34

    Nov 42 37

    Dec 54 45

    6 Sigma

  • 7/29/2019 SixSigma GB+Module

    37/50

    6 SigmaQUALITY Analyze PhaseRegression Analysis:

    Result: Regression Analysis gives following result. R-Sq indicates strength of relation.The regression equation is

    Operational Cost = 11.5 + 0.882 Inventory (Relation)Predictor Coef SE Coef T P

    Constant 11.474 2.512 4.57 0.001

    Inventory 0.88202 0.09518 9.27 0.000

    S = 3.23748 R-Sq = 89.6% R-Sq(adj) = 88.5%

    3210-1-2-3

    99

    95

    90

    80

    70

    60

    5040

    30

    20

    10

    5

    1

    Standardized Residual

    Perc

    ent

    Normal Probability Plot

    (response is Operational Cost)

    Action: Let the management fixes

    the operational cost to be 25 lakh.Equation:

    25 = 11.5 + 0.882 Inventory

    Solution:Inventory 15.3 MT

    Note: The red dots represents the

    variation of relation form the

    regression model adopted.

    6 SigmaI Ph

  • 7/29/2019 SixSigma GB+Module

    38/50

    6 SigmaQUALITY Improve PhaseStatistical Solution:

    Improve phase is all about improvement. All the previous steps are used to narrow down the

    scope of actual cause causing costly pain. Lets say that we have run the six sigma methodto improve the product quality. Later we came to know that the profile cutting machine has

    some fault. All the cut parts have slight variation. Now, what we do next? Change the

    machine or further drill on to the subject. Obviously, the later option is economically viable.

    Further, we sorted out some key factors of oxy machine that affects dimension:

    Cutting Speed. Kerf value.

    Nozzle quality.

    Plate Thickness.

    Program fault.

    But, still its quite difficult to narrow the scope of improvement. Lots of question still arises How cutting speed affects product and what damage it causes alone.What is the net impact of kerf value.

    Net effect cutting speed and kerf value.

    Do we need to change the nozzle and what will be the frequency.

    Do we need to reboot the program.

    All these answers are best addressed by Design of Experiment.

    6 SigmaI Ph

  • 7/29/2019 SixSigma GB+Module

    39/50

    6 SigmaQUALITY Improve PhaseDesign of Experiment (DOE):

    The process is always behaves in a manner of how or what the input is given and what arethe external affecting parameters. To model the behavior of each individual factor and the

    behavior in interactions with other factors, for the desired output is best done by DOE.

    In the previous example, suppose we want to figure out, how beneficial it would be if we would

    replace the old nozzles. After all no one wants useless expenses in the name of solution.Example:

    Lets say that the company wants to know the effect of cutting speed and kerf value to avoidthe costly expenses. They want to analyze the defects corresponding these factors to

    establish the standard parameters for quality product. Company conducted a test run for

    10mm thick plate for which they qualified the kerf value and cutting speed as High or Low.After conducting practical experiment, company tabulated the defect quantity as given:

    Cutting Speed (mm/m) High Low

    High 5 10

    Low 9 14

    Kerf Value (mm)

    6 SigmaI Ph

  • 7/29/2019 SixSigma GB+Module

    40/50

    6 SigmaQUALITY Improve PhaseDesign of Experiment (DOE):

    Solution:

    However, for validating the experiment they repeated the experiment.Total factors are 3 and level is 2, so total runs = 2. DOE calls for codification of High as (+)

    and Low as (-). The combination Cutting Speed x Kerf Value is coded as +, - means firstfactor is high while other is low.

    :Cutting Speed Kerf Value

    Cutting Speed x

    Kerf ValueResponse

    + + + 5 6

    + - +, - 10 11

    - + -, + 9 8

    - - + 14 12

    Main Effect : We can clearly see the impact of cutting speed and kerf value over the

    response.

    The impact is significant and this is called as main effect. However, the statistical model will

    give the relation of factors and response more appropriately.

    6 SigmaI Ph

  • 7/29/2019 SixSigma GB+Module

    41/50

    3210-1-2-3

    99

    95

    90

    80

    70

    60

    50

    40

    30

    20

    10

    5

    1

    Standardized Residual

    Percent

    Normal Probability Plot(response is Response)

    6 SigmaQUALITY Improve PhaseDesign of Experiment (DOE):

    Statistical Solution:

    When DOE test is run for the given problem, the solution is as follows:Analysis of Variance for Response, using Adjusted SS for Tests

    Source DF Seq SS Adj SS Adj MS F P

    Cutting Speed 1 15.125 15.125 15.125 17.29 0.014

    Kerf Value 1 45.125 45.125 45.125 51.57 0.002

    Cutting Speed*Kerf Value 1 0.125 0.125 0.125 0.14 0.725

    Error 4 3.500 3.500 0.875

    Total 7 63.875

    S = 0.935414 R-Sq = 94.52%R-Sq(adj) = 90.41%

    Interpretation: The p-value is 0.725 for effectof interaction, hence it is least significant for

    consideration. But p-value for cutting speed and

    kerf value is less than 0.05 so its significant.If we see the main effects plot for responses , we

    get more relation that is :

    Increase in cutting speed results decrease in

    rejections and increase in kerf value also results in

    decrease in rejections. Further, to design the

    parameter we require following regression model:

    Response = 9.3 + 1.3(Cutting Speed) 2.3(Kerf)

    6 SigmaI Ph

  • 7/29/2019 SixSigma GB+Module

    42/50

    6 SigmaQUALITY Improve PhaseDesign of Experiment (DOE): Insight

    DOE is not so easy as it seems to be in earlier example of cutting machine. DOE is not onlycrucial part of six sigma project but also serves as validating the changes for Control Phase.

    There are many ways to conduct DOE, for the sake of understanding we followed the simplest

    path. Other complex tests are used when the factors are more or it is related to the

    physiological traits for instance predicting customers buying behavior, determining

    competitive pricing, predict and optimize response rates to advertisements.Concept of DOE was laid by Ronal Fisher and is now the most vital tool in todays world.Many key figures of DOE test reveals the vital information about how the process is affected

    by several factors. Which is why, the tool is so vital in business management.

    Robust design method, also called as Taguchi method is quite vital in manufacturing

    industries. Taguchi method mainly deals with the factors on which we do not possess control

    like wear and tear of tool etc.

    This the reason why some of the manufacturing industry exceeds in process excellence over

    other.

    6 SigmaC t l Ph

  • 7/29/2019 SixSigma GB+Module

    43/50

    6 SigmaQUALITY Control PhaseStatistical Process Control (SPC):

    Improvement phase had addressed the factors causing costly pain. Control Phase is all about

    sustaining the improvement. Consider a manufacturing unit produces certain variety of partsfor instance profile, drilled and bent parts. Also, the unit is equipped with profile cuttingmachine, vertical drilling machine and press brake. If the unit makes its inspection only at final

    step i.e. before the product is delivered to the customer then three cases arises parts ok,parts to be reworked or parts to be rejected. hence the cost of production will include cost of

    ok parts, cost of defective parts and cost of rework leading to high operational cost.If we consider a system in which only the process producing parts are monitored and

    controlled, then the cost of rework, cost of rejection and cost of inspection can be drastically

    reduced. The technique is called as Statistical Process Control or SPC. SPC is all about

    managing the production process. Dr. Walter Shewhart and Dr. Edward Deming developed

    the SPC technique.SPC helps in :

    keeping production processes stable.

    predicting cause of variation before defective production.

    fixing maintenance frequency to avoid production loss.

    prevent defects.

    6 SigmaC t l Ph

  • 7/29/2019 SixSigma GB+Module

    44/50

    6 SigmaQUALITY Control PhaseStatistical Process Control (SPC): variation causes defects

    Assignable cause of variation: When the cause of variation causes differences in the output

    which is significant as per customers specification, it is called as assignable causes.Example Poor quality of RM, improper machine calibrations etc.

    Common cause of variation: When the variations are uncontrollable like weather conditions

    and causes insignificant damages to output of product, then it is called as common causes.

    They are random in nature and are predictable. However, if these causes are unpredictable

    and process is sensitive then it can affect the quality of output.

    Process monitoring tool:

    Control Charts.

    Histograms.

    Scatter Plots.

    Process Checklist.

    C & E Diagram.

    A3 Reports etc.

    These are the commonly used tools in process monitoring. But, for the sake of

    understanding we will se the Control charts only. As the others are bit complex at present.

    6 SigmaC t l Ph

  • 7/29/2019 SixSigma GB+Module

    45/50

    6 SigmaQUALITY Control PhaseControl Chart:

    Control Chart is used to capture the behavior of products over a span of time. Samples are

    randomly picked from the production line and checked for the conformance. When theprocess fails it will have an impact on the parts, which will be captured in the control chart.

    Then the variation is analyzed for immediate rectification.Elements of Control Chart:

    CTQ or X The product characteristic that we check in random inspection.

    Process Mean or The process mean of observed sample.n Sample quantity in random pick. Standard Deviation (SD) of observed X.UCL Upper Control Limit. They are 3 SD above .LCL Lower Control Limit. They are 3 SD below .

    Relation:

    = (xi) / n. = (Xi ) / n UCL = + 3 LCL = - 3 and UCL-LCL = 6 (Six Sigma Controlled)

    6 SigmaC t l Ph

  • 7/29/2019 SixSigma GB+Module

    46/50

    6 SigmaQUALITY Control PhaseControl Charts:

    WECO Rule talks about the sensitivity of variations like when 4 out of 5 dots are between

    1and 2 sigma away from mean. The rule is set by Western Electric Company.Capability Trend of 6: the six sigma way 99.7% defect free

    6 SigmaC t l Ph

  • 7/29/2019 SixSigma GB+Module

    47/50

    6 SigmaQUALITY Control PhaseX and R Chart:

    The process is about taking small sample from production line and then taking mean of it andplotting against range (R). The key question depending on DOE are

    what is the sample size.

    time to order the sampling.

    how many samples.

    The sample size depends on rational sub grouping i.e. depends on process cycle time.Further, the time to order should be in such a way that variations does not causes major

    problems. Similarly the sample size is chosen. X = Sample Mean = Xi / n

    R = X (Highest) X(Lowest)

    Center Line:

    Center Line for R chart: R = R / t, t is the number of sample.

    Center Line for X chart: X = X / t.

    -

    -

    -

    ---

    6 SigmaControl Phase

  • 7/29/2019 SixSigma GB+Module

    48/50

    6 SigmaQUALITY Control PhaseX and R Chart:

    The process is about taking small sample from production line and then taking mean of it andplotting against range (R). The key question depending on DOE are

    what is the sample size.

    time to order the sampling.

    how many samples.

    The sample size depends on rational sub grouping i.e. depends on process cycle time.Further, the time to order should be in such a way that variations does not causes major

    problems. Similarly the sample size is chosen. X = Sample Mean = Xi / n

    R = X (Highest) X(Lowest)

    Center Line: Center Line for R chart: R = R / t, t is the number of sample.

    Center Line for X chart: X = X / t.

    -

    -

    -

    ---

    6 SigmaControl Phase

  • 7/29/2019 SixSigma GB+Module

    49/50

    gQUALITY Control PhaseMistake proofing:

    Mistake-proofing is required to assure the process is self driven and unintentional variations

    made/done by any people does not affects the end result. Key features are :

    Mistake-proofing is applied in Improve and Control phase with respect to DOE.

    It is not only applied to human error but also to software or other variations.

    Standard Operating Procedure (SOP), Checklists, maintenance logs are means of control.

    FMEA enlists the possible cause of failure and is controlled on day to day basis.

    Awareness:

    All the steps discussed so far are ineffective if the quality awareness does not translates well

    down the line. Six sigma way results not only defect free processes but is also the source of

    financial savings. Majority of managers find difficulties in statistical approach because they

    trust their experience. But, six sigma digs were we fail to perceive the problem. Most of thebusiness leaders believes in increasing business, but what about sustaining the business.

    Endangered businesses situation is a result of either tough market or failure in quality as per

    customer. Six sigma process requires awareness to all the people linked to the end product.

    6 Sigma

  • 7/29/2019 SixSigma GB+Module

    50/50

    gQUALITY

    With increasing globalization, every steel company must innovate to prosper and

    compete in this new environment. POSCO was in difficult situation you might almost

    say a crisis a few years ago as we faced this new global competitive threat. As amanagement team, we felt that Six Sigma was a good vehicle to change all

    employees way of thinking, current working styles and mind-sets- Ku-taek Lee, Chairman and CEO, POSCO.

    Thank You

    By Rahil, TSPDL.Tada