Statistical Fundamentals and Control Charts

Similar documents

ISyE 512 Review for Midterm 1

Error & Uncertainty. Error. More on errors. Uncertainty. Page # The error is the difference between a TRUE value, x, and a MEASURED value, x i :

Soo King Lim Figure 1: Figure 2: Figure 3: Figure 4: Figure 5: Figure 6: Figure 7:

MOST PEOPLE WOULD RATHER LIVE WITH A PROBLEM THEY CAN'T SOLVE, THAN ACCEPT A SOLUTION THEY CAN'T UNDERSTAND.

Chapter 6 Sampling Distributions

Final Examination Solutions 17/6/2010

LINEAR REGRESSION ANALYSIS. MODULE IX Lecture Multicollinearity

Chapter 23: Inferences About Means

Statistics 511 Additional Materials

MCT242: Electronic Instrumentation Lecture 2: Instrumentation Definitions

7-1. Chapter 4. Part I. Sampling Distributions and Confidence Intervals

MATH 320: Probability and Statistics 9. Estimation and Testing of Parameters. Readings: Pruim, Chapter 4

Inferential Statistics. Inference Process. Inferential Statistics and Probability a Holistic Approach. Inference Process.

1 Models for Matched Pairs

Continuous Data that can take on any real number (time/length) based on sample data. Categorical data can only be named or categorised


II. Descriptive Statistics D. Linear Correlation and Regression. 1. Linear Correlation

Statistical Intervals for a Single Sample

Statistical Inference (Chapter 10) Statistical inference = learn about a population based on the information provided by a sample.

ENGI 4421 Confidence Intervals (Two Samples) Page 12-01

Module 1 Fundamentals in statistics

Lecture 6 Chi Square Distribution (χ 2 ) and Least Squares Fitting

Power and Type II Error

Mixed Acceptance Sampling Plans for Multiple Products Indexed by Cost of Inspection

Probability, Expectation Value and Uncertainty

Lesson 2. Projects and Hand-ins. Hypothesis testing Chaptre 3. { } x=172.0 = 3.67

Sample Size Determination (Two or More Samples)

Lecture 6 Chi Square Distribution (χ 2 ) and Least Squares Fitting

Chapter If n is odd, the median is the exact middle number If n is even, the median is the average of the two middle numbers

Estimation for Complete Data

ANALYSIS OF EXPERIMENTAL ERRORS

CHAPTER 2. Mean This is the usual arithmetic mean or average and is equal to the sum of the measurements divided by number of measurements.

CHAPTER 8 FUNDAMENTAL SAMPLING DISTRIBUTIONS AND DATA DESCRIPTIONS. 8.1 Random Sampling. 8.2 Some Important Statistics

CEE 522 Autumn Uncertainty Concepts for Geotechnical Engineering

Exam II Covers. STA 291 Lecture 19. Exam II Next Tuesday 5-7pm Memorial Hall (Same place as exam I) Makeup Exam 7:15pm 9:15pm Location CB 234

Topic 10: Introduction to Estimation

Big Picture. 5. Data, Estimates, and Models: quantifying the accuracy of estimates.

Chapter 13, Part A Analysis of Variance and Experimental Design

EECS564 Estimation, Filtering, and Detection Hwk 2 Solns. Winter p θ (z) = (2θz + 1 θ), 0 z 1

Statistical Process Control Using Two Measurement Systems

FACULTY OF MATHEMATICAL STUDIES MATHEMATICS FOR PART I ENGINEERING. Lectures

Polynomial Functions and Their Graphs

Chapter 11 Output Analysis for a Single Model. Banks, Carson, Nelson & Nicol Discrete-Event System Simulation

NO! This is not evidence in favor of ESP. We are rejecting the (null) hypothesis that the results are

Simple Linear Regression

Statistical Analysis on Uncertainty for Autocorrelated Measurements and its Applications to Key Comparisons

Properties and Hypothesis Testing

UNIT 8: INTRODUCTION TO INTERVAL ESTIMATION

Topic 9: Sampling Distributions of Estimators

Example: Find the SD of the set {x j } = {2, 4, 5, 8, 5, 11, 7}.

Linear Regression Models

Chapter 22. Comparing Two Proportions. Copyright 2010, 2007, 2004 Pearson Education, Inc.

Analysis of Experimental Data

Section 9.2. Tests About a Population Proportion 12/17/2014. Carrying Out a Significance Test H A N T. Parameters & Hypothesis

Economics 250 Assignment 1 Suggested Answers. 1. We have the following data set on the lengths (in minutes) of a sample of long-distance phone calls

1 Inferential Methods for Correlation and Regression Analysis

10. Comparative Tests among Spatial Regression Models. Here we revisit the example in Section 8.1 of estimating the mean of a normal random

Hypothesis Testing. Evaluation of Performance of Learned h. Issues. Trade-off Between Bias and Variance

Department of Civil Engineering-I.I.T. Delhi CEL 899: Environmental Risk Assessment HW5 Solution

The Method of Least Squares. To understand least squares fitting of data.

Discrete Mathematics for CS Spring 2008 David Wagner Note 22

MEASURES OF DISPERSION (VARIABILITY)

Confidence Intervals for the Population Proportion p

Lecture 24 Floods and flood frequency

NCSS Statistical Software. Tolerance Intervals

Interval Estimation (Confidence Interval = C.I.): An interval estimate of some population parameter is an interval of the form (, ),


Parameter, Statistic and Random Samples

PH 425 Quantum Measurement and Spin Winter SPINS Lab 1

Sampling Error. Chapter 6 Student Lecture Notes 6-1. Business Statistics: A Decision-Making Approach, 6e. Chapter Goals

A statistical method to determine sample size to estimate characteristic value of soil parameters

DS 100: Principles and Techniques of Data Science Date: April 13, Discussion #10

MA238 Assignment 4 Solutions (part a)

a. For each block, draw a free body diagram. Identify the source of each force in each free body diagram.

R. van Zyl 1, A.J. van der Merwe 2. Quintiles International, University of the Free State

Lecture 5: Parametric Hypothesis Testing: Comparing Means. GENOME 560, Spring 2016 Doug Fowler, GS

n but for a small sample of the population, the mean is defined as: n 2. For a lognormal distribution, the median equals the mean.

Expectation and Variance of a random variable

STA Learning Objectives. Population Proportions. Module 10 Comparing Two Proportions. Upon completing this module, you should be able to:

Topic 9: Sampling Distributions of Estimators

Chapter 6. Sampling and Estimation

Sequences. Notation. Convergence of a Sequence

Chapter 20. Comparing Two Proportions. BPS - 5th Ed. Chapter 20 1

First come, first served (FCFS) Batch

Chapter 8: Estimating with Confidence

STAT 350 Handout 19 Sampling Distribution, Central Limit Theorem (6.6)

7: Sampling Distributions

Measures of Spread: Variance and Standard Deviation

Econ 325 Notes on Point Estimator and Confidence Interval 1 By Hiro Kasahara

Mathacle. PSet Stats, Concepts In Statistics Level Number Name: Date:

Chapter 22. Comparing Two Proportions. Copyright 2010 Pearson Education, Inc.

TABLES AND FORMULAS FOR MOORE Basic Practice of Statistics

AP Statistics Review Ch. 8

Chapter 2 Descriptive Statistics

Binomial Distribution

Computing Confidence Intervals for Sample Data

Simulation. Two Rule For Inverting A Distribution Function

Continuous Functions

Wednesday s lecture. Sums of normal random variables. Some examples, n=1. Central Limit Theorem

Transcription:

Statistical Fudametals ad Cotrol Charts 1. Statistical Process Cotrol Basics Chace causes of variatio uavoidable causes of variatios Assigable causes of variatio large variatios related to machies, materials, operators or others Process i statistical cotrol the system is operatig with oly chace causes of variatio preset Process out of cotrol the system is operatig i presece of assigable causes of variatio The evetual goal of SPC is reductio or elimiatio of variability i the process by idetificatio of assigable causes 40

Basic Priciples A typical cotrol chart has cotrol limits set at values such that if the process is i cotrol, early all poits will lie betwee the upper cotrol limit (UCL) ad the lower cotrol limit (LCL). Out-of-Cotrol Situatios If at least oe poit plots beyod the cotrol limits, the process is out of cotrol If the poits behave i a systematic or oradom maer, the the process could be out of cotrol. Eample. I a electroic maufacturig process, the true = 1.5 ad stadard deviatio is = 0.15. May samples are take with each sample of size 5. The stadard deviatio of the sample average is: 0.15 0.0671 5 If the cotrol limits are set at 3 stadard deviatios from the mea, it gives the 3- sigma cotrol limits : 41

UCL = 1.5 + 3(0.0671) = 1.7013 CL= 1.5 LCL = 1.5-3(0.0671) = 1.987 The cotrol chart 4

The quality cotrol process Types of Process Variability Statioary behavior, ucorrelated data Statioary behavior, autocorrelated data Nostatioary behavior 99.7% of the Data approimately 99.7% of the data lies withi 3 of the mea (i.e., 99.7% of the data should lie withi the cotrol limits), 0.3% of the data ca fall outside 3 (or 0.3% of the data lies outside the cotrol limits). 0.007 is the probability of a Type I error or a false alarm 43

Three-Sigma Limits The use of 3-sigma limits geerally gives good results i practice. Distributio of should be ormal distributio These limits are ofte referred to as actio limits Ratioal subgroups Select cosecutive uits of productio to provide a sapshot of the process. Effective at detectig process shifts. May be ieffective i detectig if the mea has wadered out-of-cotrol ad the back Select a radom sample over the etire samplig iterval. 44

Noradom patters ca idicate out-of-cotrol coditios Cycles, treds ad rus : all above or below the ceter lie, ru up ad ru dow Rus of 8 observatios or more could idicate out-of-cotrol. A o-radom patter eample Patter is very oradom i appearace 19 of 5 poits plot below the ceter lie, while oly 6 plot above Followig 4 th poit, 5 poits i a row icrease i magitude, a ru up There is also a uusually log ru dow begiig with 18 th poit. Widely Used Cotrol Charts for Variables: -R chart ad -S chart Moitor both the mea value of the characteristic ad the variability associated with the characteristic. 45

If the process mea ad stadard deviatio are kow, we ca follow the oe phase approach to set up -R or -S cotrol charts to moitor the process o - size of the sample (sometimes called a subgroup) o i - average of the observatios i the i-th sample i=1,,3, i i1 i... o is a ormally distributed variable with mea ad stadard i deviatio o 1 is the probability that will fall betwee ad Z Z Z Z o I settig up a Shewhart cotrol chart, it typically uses Z 3. 0. / The i values will be plotted i the cotrol chart with kow : UCL= 3 Ceter Lie = LCL= 3 o We also eed to use R or S cotrol chart to moitor the process variace as productio cotiues. This will be discussed later. If the process mea ad stadard deviatio are ot kow while we ca assume that the process follows ormal or close to ormal distributio, we eed a two-phase approach to set up Shewhart cotrol chart i Phase I ad to use the 46

established cotrol charts i Phase II. The most popular oes are ad R cotrol charts. I Phase I: o m is the umber of samples selected ad o is the size of each sample o - grad average or average of the averages (this value is used as the ceter lie of the cotrol chart) 1... m m o Ri - rage of the values i the ith sample R i ma{ j ij } mi{ j ij } i,ma i,mi o R - average rage for all m samples R R1 R... Rm m cotrol Limits for the chart UCL= A R Ceter Lie = LCL= A R A is foud i Appedi VI for various values of. Cotrol Limits for the R chart UCL= D 4 R Ceter Lie = R LCL= D 3 R 47

D 3 ad D 4 are foud i Appedi VI for various values of. The above cotrol charts are based o the followig ubiased estimator of the process stadard R deviatio : ˆ as discussed i Chapter 3. d Sice A R 3 3 3 d R, so A d 3. Its value ca be foud i Appedi VI for various values of. 48

Eample R 5 R i i1 5 8.130 5 0.351 UCL= D 4 R =(.114)(0.351)=0.68749 Ceter Lie = R =0.351 LCL= D 3 R =(0)(0.351)=0 UCL= 5 i i 1 5 37.6400 5 1.5056 A R =1.5056+(0.5777) (0.351)=1.6935 Ceter Lie = = 1.5056 LCL= A R =1.5056-(0.5777) (0.351)=1.31795 49

ad R charts Cotrol limit for S Charts o S is a ubiased estimator of o S is NOT a ubiased estimator of o S is a ubiased estimator of o The stadard deviatio of S is c 4 1 c 4 o If a stadard deviatio is give, the cotrol limits for the S chart are: UCL= c 3 1 c c 3 1 c B Ceter Lie = 4 4 4 4 6 c 4 LCL= c 3 1 c c 3 1 c B 4 4 4 4 5 B 5, B 6, ad c 4 are foud i the Appedi for various values of. 50

o If a stadard deviatio is ot give, use a average sample stadard deviatio, ad the cotrol chart will be: chart whe usig S S 1 m m S i i 1 UCL= B 4 S Ceter Lie = S LCL= S The upper ad lower cotrol limits for the chart are give as UCL= A3 S Ceter Lie = LCL= A3 S S where A 3 is foud i the Appedi. ca be estimated by ˆ. c Eample 5. B 3 4 51

S 5 i1 5 5 i 5 s i i1 1850.08 5 74.001 0.351 0.0094 5 UCL= 3 A S =74.001+(1.47) (0.0094)=74.014 Ceter Lie = = 74.001 LCL= 3 A S =74.001-(1.47) (0.0094)=73.988 For S chart UCL= B 4 S =(.089) (0.0094)=0.0196 Ceter Lie = S =0.0094 LCL= B 3 S =(0) (0.0094)=0 5

3. The Shewhart Cotrol Chart for Idividual Measuremets Sample size is 1 Every uit is aalyzed The productio rate is very slow Repeat measuremets o the process differ oly because of laboratory or aalysis error. X ad Movig Rage Charts The movig rage (MR) is defied as: MRi i MR i i i 1, ad MR 1 m The X chart is the plot of the idividual observatios. The cotrol limits: UCL= MR 3 d Ceter Lie = LCL= MR 3 d The cotrol limits o the movig rage chart are: UCL= D 4 MR m Eample 5. Ceter Lie = M R LCL=0 53

MR UCL= 3 0.576 =34.088+3( ) = 35.61 d 1.18 Ceter Lie = =34.088 MR LCL= 3 0.576 =34.088-3( ) = 3.57 d 1.18 UCL= D 4 MR =3.67(0.576) = 1.871 Ceter Lie = M R = 0.576 LCL=0 54

Iterpretatio of the Charts o X charts ca be iterpreted similar to charts. MR charts caot be iterpreted the same as or R charts. o Sice the MR chart plots data that are correlated with oe aother, the lookig for patters o the chart does ot make sese. o MR chart caot really supply useful iformatio about process variability. o More emphasis should be placed o iterpretatio of the X chart. 4. Cotrol Limits, Natural Tolerace Limits ad Specificatio Limits Cotrol limits are fuctios of the atural variability of the process Natural tolerace limits represet the atural variability of the process (usually set at 3-sigma from the mea) Specificatio limits are determied by developers/desigers. There is o mathematical relatioship betwee cotrol limits ad specificatio limits. Do ot plot specificatio limits o the charts 55