Bài giảng môn Quản trị kinh doanh - Chapter 7: Quality tools:from process performance to process perfection

Chapter 7Quality Tools:From Process Performance to Process Perfection1Learning ObjectivesExplain the function of the general-purpose quality analysis tools.Explain how each quality tool aids in the QI story and DMAIC processes.Explain how statistical process control can be used to prevent defects from occurring.Calculate control limits for X-bar charts, R-charts, P-charts, and C-charts. Construct and interpret X-bar Charts, R-Charts, P-charts, and C-charts.Describe and make computations for process capability using Cp and Cpk  capability indices. Describe how acceptance sampling works and the role of the operating  characteristics curve.Explain how Six Sigma quality relates to process capability.Describe how moment-of-truth analysis can be used to improve service quality.Describe Taguchi’s quality loss function and its implications.Explain how customer relationship management systems relate to customer satisfactionDescribe how “recovery” applies to quality failures.2Quality Analysis ToolsSix Sigma’s DMAIC and TQM’s QI Story provide structure, but neither defines how activities are to be accomplished. That can be determined through the use of a broad set of analysis tools.Insert exhibit 7.1 DMAIC and QI3General-Purpose Quality Analysis ToolsProcess MapsRun ChartsCause & Effect Diagram Pareto ChartsHistogramsCheck SheetsScatter DiagramsControl Charts4General-Purpose Quality Analysis Tools: Process MapsA visual representation of a process.A Process Map for an Internet Retailer5General-Purpose Quality Analysis Tools: Run ChartsRun Charts: Plotting a variable against time.6EffectManMachineMaterialMethodEnvironmentPossible causes:The results or effect Can be used to systematically track backwards to find a possible cause of a quality problem (or effect)General-Purpose Quality Analysis Tools: Cause & Effect Diagram7General-Purpose Quality Analysis Tools: Cause & Effect DiagramAlso known as:Ishikawa DiagramsFishbone DiagramsRoot Cause Analysis8Data Analysis ExampleExhibit 7.6: SleepCheap Hotel Survey Data9Can be used to identify the frequency of quality defect occurrence and display quality performanceGeneral-Purpose Quality Analysis Tools: Histogram10General-Purpose Quality Analysis Tools: Pareto AnalysisA Variant of histogram that helps rank order quality problems so that most important can be identified50.5% of complaints are that something is dirty63.5% of complaints are about the bathroom11General-Purpose Quality Analysis Tools: Scatter Plots12General-Purpose Quality Analysis Tools: ChecksheetCan be used to keep track of defects or used to make sure people collect data in the correct mannerBilling ErrorsWrong Account Wrong AmountA/R ErrorsWrong Account Wrong AmountMonday13 Can be used to monitor ongoing production process quality and quality conformance to stated standards of qualityGeneral-Purpose Quality Analysis Tools: Control Charts14Common cause variability versus assignable cause variabilityCommon cause variability comes from random fluctuation inherent to the process.Assignable cause variability is avoidable and not part of the process.SPC takes advantage of our knowledge about the standardized distribution of these measures.Process ControlIdentifies potential problems before defects are created by watching the process unfoldIt uses X-bar Charts, R-Charts, P-charts, and C-chartsControlling Process Variability:Statistical Process Control (SPC)15Measure a sample of the process outputFour to five units of output for most applicationsMany (>25) samplesCalculate sample means ( X-bar ), grand mean (X-double bar), & ranges (R)Compare the “X-bars” being plotted to the upper and lower control limits and look for “assignable cause” variability.Assignable cause variability means that the process has changed. X-bar Chart Steps16Cp and Cpk tell us whether the process will produce defective output as part of its normal operation.i.e., is it “capable”?Control charts are maintained on an ongoing basis so that operators can ensure that a process is not changingi.e., drifting to a different level of performancei.e., is it “in control”Process Control17Measure a sample of the process outputFour to five units of output for most applicationsMany (>25) samplesCalculate sample means ( X ), grand mean (X), & ranges (R)Calculate “process capability”Can you deliver within tolerances defined by the customerTraditional standard is “correct 99.74% of the time”Monitor “process control”Is anything changing about the process?In terms of mean or variationSPC Steps18X-Bar and R-Chart ConstructionInsert Exhibit 7.1719 Steps:Calculate Upper & Lower Control Limits (UCL & LCL).Use special charts based on sample sizePlot X-bar value for each sampleInvestigate “Nonrandom” patternsControl Charts: X-barExhibit 7.18 X-bar Chart for Example 7.2Distinguishing between random fluctuation and fluctuation due to an assignable cause. X-bar chart tracks the trend in sample means to see if any disturbing patterns emerge.????20Nonrandom Patterns on Control ChartsInvestigate the process if X-bar or R chart illustrates:One data point above +3 or below -39 points in a row, all above or all below the mean6 points in a row, all increasing or all decreasing14 points in a row alternating up and down4 out of 5 points in a row in Zone B or beyond15 points in a row in Zone 3, above or below the center line 8 points in a row in Zone B, A, or beyond, on either side of the center line with no points in Zone C21 R-charts R-charts monitor variation within each sample.R-charts are always used with X-bar charts.Exhibit 7.22 R-Chart for Example 7.4StepsCalculate Upper & Lower Control Limits (UCL & LCL).Use special tables based on sample size.Plot the R value for each sampleInvestigate “Nonrandom” patterns????22Process CapabilityCapability Index: Quantifies the relationship between control limits and customer specifications.A process is “capable” when all of the common cause variability occurs within the customer’s specification limits.Cp is used to determine “capability” when the process is centered.23Cp CalculationFor Centered ProcessesCp compares the range of the customer’s expectations to the range of the process to make sure that all common cause variability is inside of the customer’s specifications. Cp = UCS - LCS 6σUCS - Upper control specificationLCS - Lower control specification - Standard deviation of process performanceIf Cp > 1.000 the process is considered capable.`24Example 7.4: Cp CalculationCustomer specificationMean of .375 inches+ or - .002 inchesTherefore, customer specification limits at .373 and .377Process performanceActual mean is .375Standard deviation is 0.0024Cp = 0.377 – 0.373 6(0.0024) = 0.27778 The process is not capable.25Process Capability for Uncentered ProcessesSome processes are intentionally allowed to “shift.”In these cases, the range of process variability moves toward one of the customer specifications as the process shifts. 26Process Capability for Uncentered ProcessesAs soon as one of the “tails” of the process distribution crosses the customer specifications, the process is no longer capable27Process Capability for Uncentered ProcessesThe process capability index for uncentered processes checks both ends of the distribution to ensure that the process has not shifted beyond the customer specifications.28Cpk CalculationLCS - Lower control specificationUCS - Upper control specificationX - “Grand” mean of process performance - Standard deviation of process performanceIf Cpk is > 1.000 then the process is “Capable”Translation, we will produce good parts at least 99.74% of the time29Example 7.3: Cpk CalculationCustomer specificationMean of .375 inches+ or - .002 inchesTherefore, customer specification limits at .373 and .377Process performanceActual mean is .376Standard deviation is 0.0003Cpk = min[ 0.376 – 0.373 , 0.377 – 0.376 ] 0.0009 0.0009= min [3.333, 1.111]= 1.111The process is capable.30Process Control Charts for AttributesP-chartsUsed to monitor the proportion or percentage of items defective in a given sample.UCL=LCL=n = the sample size = the long-run average and center lineZ is the number of normal standard deviations for the desired confidence31Process Control Charts for AttributesC-chartsUsed to monitor the counts of noncomformities per unit.UCL = LCL =32Acceptance SamplingPurposesSampling to accept or reject the immediate lot of product at handEnsure quality is within predetermined levelAdvantagesDisadvantages-Economy-Less handling damage-Fewer inspectors-Upgrading of the inspection job-Applicability to destructive testing-Entire lot rejection (motivation for improvement) -Risks of accepting “bad” lots and rejecting “good” lots-Added planning and documentation-Sample provides less information than 100-percent inspection33Acceptance SamplingAcceptable Quality Level (AQL)Is the max. acceptable percentage of defectives that defines a “good” lotProducer’s risk is the probability of rejecting a good lotLot tolerance percent defective (LTPD)Is the percentage of defectives that defines consumer’s rejection pointConsumer’s risk is the probability of accepting a bad lotThe sampling plan is developed based on risk tolerance to determine size of sample and number in sample that can be defectiveExhibit 7.26 Operating Characteristics Curve34Six Sigma QualityExhibit 7.28 Process Capability for Six Sigma Quality“Six sigma” refers to the variation that exists within plus or minus six standard deviations of the process outputs35Six Sigma QualityIn “process capability” terms, Six Sigma means that control limits set at plus or minus 6 σ will be inside of the customer’s specifications.This greatly reduces the likelihood of a defect occurring from common cause variability.36Six Sigma Quality – Role of interdependencies6 is often needed when products are complex.At 3 quality, for example, the probability that an assembly of interdependent parts works, given “n” parts and the need for all parts to work:1 part = .99741 = 99.74%10 parts = .997410 = 97.43%50 parts = .997450 = 87.79%100 parts = .9974100 = 77.08%267 parts = .9974267 = 49.90% 1000 parts = .99741000 = 7.40%37The odds of common cause variability creating a result that is 6 from the mean are 2 in 1 billion99.9999998% confident of a good outcomeIn practice, process mean is allowed to shift ±1.5 Six Sigma and Failure Rates38Taguchi MethodAs deviation from the target increases, customers get increasingly dissatisfied and costs increase.Traditional views define deviation in terms of being “good” or “defective.” Taguchi views deviation in terms of costs that occur even if the deviation is slight, and increasing costs as deviation increases.39Moment-of-Truth Analysis: The identification of the critical instances when a customer judges service quality and determines the experience enhancers, standard expectations, and experience detractors.Experience enhancers: Experiences that make the customer feel good about the interaction and make the interaction better.Standard expectations: Experiences that are expected and taken for granted.Experience detractors: Experiences viewed by the customer as reducing the quality of service.Moment-of-Truth Analysis40RecoveryThere will always be times when customers do not get what they want.Failure to meet customers’ expectations does not have to mean lost customers.Recovery plans: Policies for how employees are to deal with quality failures so that customers will return.Example: A recovery for a customer who has had a bad meal at a restaurant might include eliminating the charges for the meal, apologizing, and offering gift certificates for future meals.41