Quality Control McGraw-Hill/Irwin Copyright © 2012 by The McGraw-Hill Companies, Inc. All rights reserved. You should be able to: 1. List and briefly explain the elements in the control process 2. Explain how control charts are used to monitor a process, and the concepts that underlie their use 3. Use and interpret control charts 4. Perform run tests to check for nonrandomness in process output 5. Assess process capability Instructor Slides 10-2 Quality Control A process that evaluates output relative to a standard and takes corrective action when output doesn’t meet standards If results are acceptable no further action is required Unacceptable results call for correction action Instructor Slides 10-3 Instructor Slides 10-4 Inspection An appraisal activity that compares goods or services to a standard Inspection issues: 1. 2. 3. 4. Instructor Slides How much to inspect and how often At what points in the process to inspect Whether to inspect in a centralized or on-site location Whether to inspect attributes or variables 10-5 Figure 10.2 How Much/How Often Where/When Centralized vs. On-site Inputs Acceptance sampling Transformation Outputs Acceptance sampling Process control 10-6 Instructor Slides 10-7 University: Is each student achieving acceptable performance? Chemistry: Does each product contain the correct amount of chemical? Automobile: Does the automobile performs as intended? Bank: Are the transaction completed with precision? 10-8 There are two general approaches to quality management. The inspection approach (inspects the quality of input and the finished product) The detection approach (focuses on the production process, and detects when it starts going out of control) 10-9 There are two general approaches to quality management. The inspection approach (inspects the quality of input and the finished product) The detection approach (focuses on the production process, and detects when it starts going out of control) 10-10 Typical Inspection Points: Raw materials and purchased parts Finished products Before a costly operation Before an irreversible process Before a covering process Instructor Slides 10-11 Effects on cost and level of disruption are a major issue in selecting centralized vs. on-site inspection Centralized Specialized tests that may best be completed in a lab More specialized testing equipment More favorable testing environment On-Site Quicker decisions are rendered Avoid introduction of extraneous factors Quality at the source Instructor Slides 10-12 Table 10.1 Type of business Fast Food Inspection points Cashier Counter area Eating area Building Kitchen Hotel/motel Parking lot Accounting Building Main desk Supermarket Cashiers Deliveries Characteristics Accuracy Appearance, productivity Cleanliness Appearance Health regulations Safe, well lighted Accuracy, timeliness Appearance, safety Waiting times Accuracy, courtesy Quality,10-13 quantity 10-14 Quality control seeks Quality of Conformance A product or service conforms to specifications A tool used to help in this process: SPC Statistical evaluation of the output of a process Helps us to decide if a process is “in control” or if corrective action is needed Statistical Process Control (SPC) is a form of hypothesis testing employed in industry. Its objective is to monitor the quality of a firm’s products and services Instructor Slides 10-15 Two basic questions: concerning variability: 1. Issue of Process Control Are the variations random? If nonrandom variation is present, the process is said to be unstable. 2. Issue of Process Capability Given a stable process, is the inherent variability of the process within a range that conforms to performance criteria? Instructor Slides 10-16 Variation Random (common cause) variation: Natural variation in the output of a process, created by countless minor factors Assignable (special cause) variation: A variation whose cause can be identified. A nonrandom variation Instructor Slides 10-17 SPC involves periodically taking samples of process output and computing sample statistics: Sample means The number of occurrences of some outcome Sample statistics are used to judge the randomness of process variation Instructor Slides 10-18 Sampling Distribution A theoretical distribution that describes the random variability of sample statistics The normal distribution is commonly used for this purpose Central Limit Theorem The distribution of sample averages tends to be normal regardless of the shape of the process distribution Instructor Slides 10-19 Instructor Slides 10-20 Sampling and corrective action are only a part of the control process Steps required for effective control: Define: What is to be controlled? Measure: How will measurement be accomplished? Compare: There must be a standard of comparison Evaluate: Establish a definition of out of control Correct: Uncover the cause of nonrandom variability and fix it Monitor: Verify that the problem has been eliminated Instructor Slides 10-21 Control Chart A time ordered plot of representative sample statistics obtained from an ongoing process (e.g. sample means), used to distinguish between random and nonrandom variability Control limits The dividing lines between random and nonrandom deviations from the mean of the distribution Upper and lower control limits define the range of acceptable variation Instructor Slides 10-22 Each point on the control chart represents a sample of n observations Instructor Slides 10-23 The essence of statistical process control is to assure that the output of a process is random so that future output will be random. 10-24 UCL CL LCL 0 1 2 3 4 5 6 7 8 9 101112131415161718192021222324252627 Sample 10-25 Variations and Control Random variation: Natural variations in the output of a process, created by countless minor factors Assignable variation: A variation whose source can be identified 10-26 All products exhibit some degree of variation. There are two sources of variation Random or chance variation: Natural variations in the output of process, created by countless minor factors that cannot be eliminated without changing the process Assignable variation: A variation whose source can be identified and eliminated 10-27 Figure 10.5 μ = E(X) Sampling distribution s = σ/√n Process distribution Mean 10-28 Figure 10.6 Standard deviation Mean 95.44% 99.74% 10-29 Figure 10.7 Sampling distribution Process distribution Mean Lower control limit Upper control limit 10-30 Type I error Concluding a process is not in control when it actually is. The probability of rejecting the null hypothesis when the null hypothesis is true. Manufacturer’s Risk Type II error Concluding a process is in control when it is not. The probability of failing to reject the null hypothesis when the null hypothesis is false. Consumer’s Risk Instructor Slides 10-31 Table 10.2 In control Out of control In control No Error Out of control Type II Error (consumers risk) Type I error (producers risk) No error 10-32 Instructor Slides 10-33 Instructor Slides 10-34 Variables generate data that are measured Mean control charts Used to monitor the central tendency of a process. “x- bar” charts Range control charts Used to monitor the process dispersion R charts Instructor Slides 10-35 UCL CL LCL 0 1 2 3 4 5 6 7 8 9 101112131415161718192021222324252627 Sample 10-36 k k x xi i 1 R R i 1 k where x Average of sample means x i mean of sample i k number of samples Instructor Slides i k where R Average of sample ranges Ri Range of sample i 10-37 The x chart This chart is employed to determine whether the distribution mean has changed. Upper control limit x Lower control limit Start with k samples when the process is under control. Average the k sample averages. Average the k sample ranges. 10-38 X UCL = LCL = x za 2 x za 2 n n 10-39 Given the following samples which have been collected from the length of time to process loan application: Sample 1 Observation Mean Range 2 3 4 5 1 12.11 12.15 12.09 12.12 12.09 2 12.10 12.12 12.09 12.10 12.14 3 12.11 12.10 12.11 12.08 12.13 4 12.08 12.11 12.15 12.10 12.12 12.10 12.12 12.11 12.10 12.12 0.03 0.05 0.06 0.04 0.05 10-40 Q1: If the standard deviation of the process is σ = .02, find the three-sigma (z = 3) control limits. Grand Mean = (12.10 + 12.12 + 12.11 + 12.10 + 12.12)/5 = 12.11 UCL = 12.11 + (3)(.02 / 2) = 12.14 LCL = 12.11 - (3)(.02 /2) = 12.08 X 10-41 Used to monitor the central tendency of a process x chart Control Limits UCLx x A2 R LCLx x A2 R where A2 a control chart factor based on sample size, n Instructor Slides 10-42 Q2: Find the mean control limits assuming that the standard deviation is not known: R = (.03+.05+.06+.04+.05) / 5 = .046 UCL = 12.11 + (.73)(.046) = 12.14358 LCL = 12.11 – (.73)(.046) = 12.07642 10-43 Used to monitor process dispersion R Chart Control Limits UCLR D4 R LCLR D3 R where D3 a control chart factor based on sample size, n D4 a control chart factor based on sample size, n Instructor Slides 10-44 Range chart UCL = (2.28)(.046) = .10488 LCL =(0)(.046) = 0 10-45 Instructor Slides 10-46 Figure 10.10B Sampling Distribution (process variability is increasing) UCL Does not reveal increase x-Chart LCL UCL R-chart Reveals increase LCL 10-47 To determine initial control limits: Obtain 20 to 25 samples Compute appropriate sample statistics Establish preliminary control limits Determine if any points fall outside of the control limits If you find no out-of-control signals, assume the process is in control If you find an out-of-control signal, search for and correct the assignable cause of variation Resume the process and collect another set of observations on which to base control limits Plot the data on the control chart and check for out-of-control signals Instructor Slides 10-48 Attributes generate data that are counted. p-Chart Control chart used to monitor the proportion of defectives in a process c-Chart Control chart used to monitor the number of defects per unit Instructor Slides 10-49 When observations can be placed into two categories. Good or bad Pass or fail Operate or don’t operate When the data consists of multiple samples of several observations each Instructor Slides 10-50 Total number of defectives p Total number of observatio ns p (1 p ) ˆ p n UCL p p z (ˆ p ) LCL p p z (ˆ p ) Instructor Slides 10-51 Using 4 samples of 100 credit card statements Sample # each, an auditor found 1 the following: 2 Total # with Error 4 3 3 5 4 8 20 10-52 The fraction defective estimate is: = 20 / (4 x 100) = .05 Q: What is the probability distribution for this experiment? The standard deviation of fraction defective with sample size n = 100 is: P σp = p (1 p ) n 10-53 σp = (.05)(.95) / 100 σp = .02179 Note that n = 100 is the sample size and not the number of samples 10-54 Control limits are computed using the following formulas: UCL = P + Z σp LCL = P - Z σp For example Z =3 results in the following control limits: UCL = .05 + (3)(.02179) = .11537 LCL = .05 – (3)(.02179) = .01537 10-55 Sample # with Fraction # Error Defective 1 4 .04 In control? Yes 2 3 .03 Yes 3 5 .05 Yes 4 8 .08 Yes 10-56 For Z= 3, the type I error for example 2 is calculated from the Normal probability table: α = 1 –(.4987 x 2) = .0026 10-57 Use only when the number of occurrences per unit of measure can be counted; non-occurrences cannot be counted. Scratches, chips, dents, or errors per item Cracks or faults per unit of distance Breaks or Tears per unit of area Bacteria or pollutants per unit of volume Calls, complaints, failures per unit of time UCL c c z c LCL c c z c Instructor Slides 10-58 The following table displays the number of scratches in the exterior paint of 4 samples of ten automobiles each: Sample # Total Number 1 35 2 42 3 28 4 45 150 10-59 The average number of defects is: c = 150 / (4 x 10) = 3.75 Q: What is the probability distribution for this experiment? The standard deviation is: σc = c σc = 1.9365 10-60 Control limits are computed using the following formulas: UCL = c c +Z c c LCL = -Z For example Z =3 results in the following control limits: UCL = 3.75 + (3)(1.9365) = 9.5595 LCL = 3.75 – (3)(1.9365) = -2.0595 or 0 10-61 At what points in the process to use control charts What size samples to take What type of control chart to use Variables Attributes Instructor Slides 10-62 Even if a process appears to be in control, the data may still not reflect a random process Analysts often supplement control charts with a run test Run test A test for patterns in a sequence Run Sequence of observations with a certain characteristic Instructor Slides 10-63 Instructor Slides 10-64 Run test – a test for randomness Any sort of pattern in the data would suggest a non-random process All points are within the control limits - the process may not be random 10-65 Trend Cycles Bias Mean shift Too much dispersion 10-66 Counting Runs Figure 10.12 Counting Above/Below Median Runs B A Figure 10.13 A B A B B B A Counting Up/Down Runs U U D U (7 runs) A B (8 runs) D U D U 10-67 U D Statistical process control is based on repetitive sampling procedures and hypotheses tests. Therefore, we can run tests to detect an out of control state based on patterns developed by the repeated measurements. We may conclude that a process is out of control when certain patterns of the process behavior are detected, even if no observation falls outside the control limits. 10-68 Above/Below Median runs: E(r)med = N/2 + 1 med N 1 4 u/d 16 N 29 90 Up/Down runs: E(r)u/d = (2N – 1) /3 10-69 Text example 6, pages 468-469 10-70 Once a process has been determined to be stable, it is necessary to determine if the process is capable of producing output that is within an acceptable range Tolerances or specifications Range of acceptable values established by engineering design or customer requirements Process variability Natural or inherent variability in a process Process capability The inherent variability of process output (process width) relative to the variation allowed by the design specification (specification width) Instructor Slides 10-71 Lower Upper Specification Specification Process variability (width) exceeds specifications Lower Specification Lower Specification Upper Specification Process variability (width) matches specifications width Upper Specification Process variability (width) is less than the specification width Instructor Slides 10-72 UTL - LTL Cp 6 where UTL upper tole rance (specifica tion) limit LTL lower tole rance(spec ification) limit Instructor Slides 10-73 Used when a process is not centered at its target, or nominal, value C pk min C pu , C pl UTL x x LTL min , 3 3 Instructor Slides 10-74 The control chart is divided into 6 horizontal zones. Example: Zones for the X control chart. Upper control limit A B C C B A lower control limit x 3S / n x 2S / n x S/ n x x S/ n x 2S / n x 3S / n – Several pattern-tests can be applied to these zones. Each one represents a rare event, which if it occurs indicates 10-75 an out of control state. A B C C B A One point beyond zone A. A A B B C C 9 2 4 8 C C 6 B 1 B 3 5 7 A A Nine points in a row in zone C or beyond (on same side of CL). 5 1 2 3 4 Six increasing (decreasing) points in a row. 6 A B C C B A Fourteen points in a row alternating up and down. A B C C B A A B C C B A A B C C B A A B C C B A 2 out of 3 points in a row in zone A or beyond (same side of CL). 4 out of 5 points in a row in zone B or beyond. 15 points in a row in zone C (on both sides of CL). 10-76 8 points in a row beyond zone C (on both sides of CL). The proportion of defectives in a computer disks production line has to be monitored. The experiment Every hour a random sample of 200 disks is taken. Each disk is tested to determine if it is defective. Based on the first 40 samples, draw a p chart and determine whether the process was out of control. 10-77 Solution Sample Defective Proportion 1 2 3 19 5 16 .095 .025 .080 Mean sample proportion = p = (.095+..25+.080+…)/40 =.05762 ..……………………………... Centerline p .05762 Lower control lim it p 3 p (1 p ) .05762(1 .05762) .05762 3 .0088188 n 200 p (1 p ) .05762(1 .05762) Upper control lim it p 3 .05762 3 .1071 n 200 An estimate of the s tan dard deviation p(1 p ) .01648 n 10-78 The zones P-Chart .05762+3(.01648) =.1071 .05762+2(.01648) =.09057 .05762+(.01648) =.0741 0.12 0.1 Zone A 0.08 Zone B 0.06 Zone C Zone C 0.04 Zone B 0.02 Zone A 0 0 1 2 3 4 5 6 7 8 9101112131415161718192021222324252627282930313233343536373839404142 There is no evidence to infer that the process was out of control 10-79 Managers should have response plans to investigate cause May be false alarm (Type I error) May be assignable variation 10-80 Simplify Standardize Mistake-proof Upgrade equipment Automate Instructor Slides 10-81 Quality is a primary consideration for nearly all customers Achieving and maintaining quality standards is of strategic importance to all business organizations Product and service design Increase capability in order to move from extensive use of control charts and inspection to achieve desired quality outcomes Instructor Slides 10-82