P. 1
8. ReliabilityandSurvivalAnalysisAllTopics[1][1]

8. ReliabilityandSurvivalAnalysisAllTopics[1][1]

|Views: 351|Likes:
Published by Zaid Tariq Alabiry

More info:

Published by: Zaid Tariq Alabiry on Feb 05, 2012
Copyright:Attribution Non-commercial

Availability:

Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less

02/11/2013

pdf

text

original

Sections

  • Test Plans
  • Test Plans Overview
  • Failure Censoring
  • Time Censoring
  • Type I and Type II Errors
  • Demonstration Test Plans
  • Estimation Test Plans
  • Accelerated Life Test Plans
  • Distribution Analysis
  • Distribution Analysis Overview
  • Estimation methods
  • Distribution Analysis Data
  • Goodness-of-fit statistics
  • Stacked vs. Unstacked data
  • Arbitrarily Censored Data
  • Right Censored Data
  • Growth Curves
  • Growth Curve Overview
  • Data - Growth Curves
  • Growth curves - exact data
  • Growth curves - interval data
  • Growth curves - grouped interval data
  • Using Cost or Frequency Columns
  • Using Time and Retirement Columns
  • Parametric Growth Curve
  • Nonparametric Growth Curve
  • Accelerated Life Testing
  • Regression with Life Data Overview
  • Worksheet Structure for Regression with Life Data
  • To perform accelerated life testing with uncensored/right censored data
  • To perform accelerated life testing with uncensored/arbitrarily censored data
  • Transforming the accelerating variable
  • Percentiles and survival probabilities
  • Accelerated Life Testing - Censor
  • Accelerated Life Testing - Estimate
  • To estimate percentiles and survival probabilities
  • Accelerated Life Testing - Graphs
  • To modify the relation plot
  • Relation plot
  • Probability plot for each accelerating level based on fitted model
  • Probability plots
  • Accelerated Life Testing - Options
  • Accelerated Life Testing - Results
  • Accelerated Life Testing - Storage
  • Example of Accelerated Life Testing
  • Output
  • Regression with Life Data
  • Data - Regression with Life Data
  • Uncensored/arbitrarily censored data
  • Uncensored/right censored data
  • Failure times
  • To perform regression with uncensored/right censored data
  • To perform regression with uncensored/arbitrarily censored data
  • Estimating the model parameters
  • Factor variables and reference levels
  • Multiple degrees of freedom test
  • Regression with Life Data - Censor
  • Regression with Life Data - Estimate
  • Regression with Life Data - Graphs
  • Probability plots for regression with life data
  • To draw a probability plot of the residuals
  • Regression with Life Data - Options
  • To control estimation of the parameters
  • To change the reference factor level
  • Regression with Life Data - Results
  • To perform multiple degrees of freedom tests
  • Regression with Life Data - Storage
  • Example of Regression with Life Data
  • Default output
  • Probit Analysis
  • Probit Analysis Overview
  • Data - Probit Analysis
  • To perform a probit analysis
  • Probit model and distribution function
  • Natural response rate
  • Percentiles
  • Survival and cumulative probabilities
  • Probit Analysis - Estimate
  • To request survival probabilities
  • Probit Analysis - Graphs
  • To draw a survival plot
  • Survival plots
  • Probit Analysis - Options
  • Probit Analysis - Results
  • To modify the table of percentiles
  • Probit Analysis - Storage
  • Example of a Probit Analysis
  • Probit Analysis - Output
  • References - Reliability and Survival Analysis
  • Index

Reliability and Survivial Analysis

Table Of Contents

Table Of Contents
Test Plans .............................................................................................................................................................................. 7 Test Plans Overview......................................................................................................................................................... 7 Failure Censoring ............................................................................................................................................................. 7 Time Censoring ................................................................................................................................................................ 8 Type I and Type II Errors.................................................................................................................................................. 8 Demonstration Test Plans ................................................................................................................................................ 8 Estimation Test Plans..................................................................................................................................................... 12 Accelerated Life Test Plans............................................................................................................................................ 16 Distribution Analysis............................................................................................................................................................. 23 Distribution Analysis Overview ....................................................................................................................................... 23 Estimation methods ........................................................................................................................................................ 23 Distribution Analysis Data............................................................................................................................................... 24 Goodness-of-fit statistics ................................................................................................................................................ 24 Stacked vs. Unstacked data ........................................................................................................................................... 25 Arbitrarily Censored Data ............................................................................................................................................... 25 Right Censored Data ...................................................................................................................................................... 64 Growth Curves ................................................................................................................................................................... 115 Growth Curve Overview ............................................................................................................................................... 115 Data - Growth Curves................................................................................................................................................... 115 Growth curves - exact data........................................................................................................................................... 115 Growth curves - interval data ....................................................................................................................................... 116 Growth curves - grouped interval data ......................................................................................................................... 117 Using Cost or Frequency Columns .............................................................................................................................. 118 Using Time and Retirement Columns .......................................................................................................................... 118 Parametric Growth Curve ............................................................................................................................................. 118 Nonparametric Growth Curve....................................................................................................................................... 130 Accelerated Life Testing .................................................................................................................................................... 141 Regression with Life Data Overview ............................................................................................................................ 141 Accelerated Life Testing ............................................................................................................................................... 141 Worksheet Structure for Regression with Life Data ..................................................................................................... 142 To perform accelerated life testing with uncensored/right censored data .................................................................... 142 To perform accelerated life testing with uncensored/arbitrarily censored data ............................................................ 143 Transforming the accelerating variable ........................................................................................................................ 143 Percentiles and survival probabilities ........................................................................................................................... 144 Accelerated Life Testing - Censor ................................................................................................................................ 144 Accelerated Life Testing - Estimate.............................................................................................................................. 144 To estimate percentiles and survival probabilities........................................................................................................ 145 Accelerated Life Testing - Graphs ................................................................................................................................ 145 To modify the relation plot ............................................................................................................................................ 145 Relation plot.................................................................................................................................................................. 146 Probability plot for each accelerating level based on fitted model................................................................................ 146 Probability plots ............................................................................................................................................................ 146 Accelerated Life Testing - Options ............................................................................................................................... 147 Accelerated Life Testing - Results................................................................................................................................ 147 Accelerated Life Testing - Storage ............................................................................................................................... 147 Example of Accelerated Life Testing ............................................................................................................................ 148

Copyright © 2003–2005 Minitab Inc. All rights reserved.

3

.............................................................................................................................. 156 Probability plots for regression with life data .................................................................. 168 To modify the table of percentiles ........................................... 168 Probit Analysis ......... 157 To change the reference factor level ............................................................................................................... 155 Regression with Life Data ................................................................................................................................................................................................... 164 Probit model and distribution function ....................................... 163 Probit Analysis...................................... 157 Regression with Life Data ........................... 156 To draw a probability plot of the residuals........................................................... 157 To perform multiple degrees of freedom tests.................................................................................................................................................................................................................................................................. 163 Data ........ 153 To perform regression with uncensored/right censored data ........................................................................................................................................................................................................................ 163 To perform a probit analysis .............................................................................................................................................................................................................Storage ........................................................ 167 Survival plots ... 168 To control estimation of the parameters ..............................................................................................................................Storage...... 154 Multiple degrees of freedom test ................Options ...................................................................................................................................................................... .................................................................................................................................................................................................................. 154 Estimating the model parameters...................................................... All rights reserved.....................................................................................................................................................................................Probit Analysis ............................................................................................................................................ 150 Regression with Life Data .................Reliability and Survivial Analysis Output............................................................................. 165 Factor variables and reference levels ....................................... 154 Factor variables and reference levels ................................................................................................................................................................................................. 155 To estimate percentiles and survival probabilities............................................... 157 To control estimation of the parameters ......................................................................................................................................................................................................................... 164 Estimating the model parameters.........................Options ............................................................................................................................... 166 To request survival probabilities .............. 155 Regression with Life Data ..................................................................................... 165 Natural response rate .................................................................................................................................................. 151 Data ................................................................................................................................................................. 168 Probit Analysis ........... 163 Probit Analysis Overview............................................................................................................................................................................................................... 158 Regression with Life Data .............................................. 165 Percentiles................................................................................................................................................ 169 4 Copyright © 2003–2005 Minitab Inc.........................................................................Graphs................................................................................Results ................................................................................................Regression with Life Data...................................................................................................... 152 Uncensored/right censored data ............................................... 156 Regression with Life Data .............................. 151 Regression with Life Data Overview ......................................... 167 To draw a survival plot......................................... 158 Example of Regression with Life Data .................................................................................................................................................................................................................................................................................................................................................................................................................... 156 Regression with Life Data ......................................................................................................................................................................................................................................Censor................................. 167 Probability plots .............................................................. 161 Probit Analysis .................... 151 Regression with Life Data .................................................................................................................................................................................................................................................... 158 Default output .........................................................................................................................................................................................................................Results............................................................................................................ 167 Probit Analysis .................... 169 Probit Analysis ............................................................................................................................................................ 166 Probit Analysis ............................................................................................................................................................................................................ 166 Survival and cumulative probabilities ...... 153 Failure times ............Estimate .............. 154 To perform regression with uncensored/arbitrarily censored data ......................................................Estimate.........................................................................................................................................................................................................Graphs ........... 151 Uncensored/arbitrarily censored data ......................................................................................

..........................................Reliability and Survival Analysis........................ 170 Probit Analysis ............................................................................... 5 ................. 173 References .....................................................................................................................................................................................................Output............................................................ All rights reserved..................................................Table Of Contents Example of a Probit Analysis..................................... 177 Copyright © 2003–2005 Minitab Inc..................................................................... 175 Index ...

.

Test Plans

Test Plans
Test Plans Overview
Use Minitab's test planning commands to determine the sample size and testing time needed to estimate model parameters or to demonstrate that you have met specified reliability requirements. A test plan includes: • • • The number of units you need to test A stopping rule − the amount of time you must test each unit or the number of failures that must occur Success criterion − the number of failures allowed while the test still passes (for example, every unit runs for the specified amount of time and there are no failures)

Three kinds of test plans are available: demonstration, estimation, and accelerated life.

Demonstration test plans
Use demonstration test plans to determine the sample size or testing time needed to demonstrate, with some level of confidence, that the reliability exceeds a given standard. There are two types of demonstration tests: • Substantiation tests provide statistical evidence that a redesigned system has suppressed or significantly reduced a known cause of failure. You are testing: H0: The redesigned system is no different from the old system. H1: The redesigned system is better than the old system. • Reliability tests provide statistical basis that a reliability specification has been achieved. You are testing: H0: The system reliability is less than or equal to a goal value. H1: The system reliability is greater than a goal value. You can rewrite these hypotheses in terms of the scale (Weibull or exponential distributions) or location (other distributions), a percentile, the reliability at a particular time, or the mean time to failure (MTTF). For example, you can test whether or not the MTTF for a redesigned system is greater than the MTTF for the old system. Minitab provides an m-failure test plan for substantiation and reliability testing. If more than m failures occur in an m-failure test, the test fails.

Estimation test plans
Use estimation test plans to determine the number of test units that you need to estimate percentiles or reliabilities with a specified degree of precision. Estimation test plans are similar to classical sample-size problems, but computations are more intensive because the data are usually censored. Use estimation test plans to answer questions such as: • • How many units must I test to estimate the 10th percentile with a 95% lower confidence bound within 100 hours of the estimate? How long must I run the test to estimate the reliability at 500 hours with a 95% lower confidence bound within 0.05 of the estimate?

Accelerated life test plans
Use accelerated life test plans to determine the number of units to test and how to allocate those units across stress levels for an accelerated life test or to determine the standard error for the parameter you wish to estimate given a fixed number of test units. Use accelerated life test plans to answer questions such as: • • • How many units must I test to estimate the 10th percentile with a 95% upper confidence bound within 100 hours of the estimate? What is the best allocation of 20 units across 3 stress levels in order to estimate the reliability at 1000 hours? Twenty units are available for testing. What standard error can you expect for the estimate of the 500-hour reliability?

To obtain an accelerated test plan, you provide the stress values and, optionally, the proportionate allocation of test units. Minitab evaluates the resulting plans and displays the "best" plans with respect to minimizing the variance.

Failure Censoring
Failure censoring is useful for: • Testing lower percentiles − For any percentile, increasing the test duration improves the precision of your estimate. However, you will see little improvement in precision when you run a test far beyond the estimated percentile. For example, if you estimate the 10th percentile, you obtain important gains in precision by running the test until around

Copyright © 2003–2005 Minitab Inc. All rights reserved.

7

Reliability and Survivial Analysis

15% of the units fail, but little improvement by running the test longer. In fact, running the test beyond 15% of the units failing could bias your estimate of the 10th percentile. • Replacing test units − If you have a limited number of test positions, you can use failure censoring to determine when to replace unfailed units. For example, if you want to estimate the 10th percentile, but can only test 5 units at a time, you may want to replace all 5 units after the first failure in each group. In this case, you are failure-censoring when 20% of the units in each group have failed.

Time Censoring
Testing all units to failure in a life test usually does not make sense, especially if you are only interested in the lower percentiles of the distribution. For any percentile of interest, the precision of your results depends on: • • Test duration Sample size

To minimize cost, you need to balance the test duration and sample size. For a given precision, Minitab displays a list of sample sizes for each censoring time you provide. As time increases, the sample size decreases. Choose the time and sample size combination that minimizes costs. For an accelerated life test plan, you only need to provide one set of censor times. Each time in the set corresponds to the censor time at a stress level. The first time corresponds to the lowest stress level, the second time corresponds to the second stress level, and so on.

Type I and Type II Errors
Hypothesis tests have four possible outcomes: Null Hypothesis (H0) Decision True False Type II error p=β Correct decision p=1−β Fail to reject H0: Correct decision p=1−α Reject H0: Type I error p=α

The outcome of the test depends on whether the null hypothesis (H0) is true or false and whether you reject or fail to reject it. • When H0 is true and you reject it, you make a Type I error. The probability (p) of making a Type I error is called alpha (α), or the level of significance of the test. • When H0 is false and you fail to reject it, you make a Type II error. The probability (p) of making a Type II error is called beta (β). The power of a test is the probability of correctly rejecting H0 when it is false. In other words, power is the likelihood that you will identify a significant effect when one exists.

Demonstration Test Plans
Demonstration Test Plans
Stat > Reliability/Survival > Demonstration Test Plans Use to demonstrate that you have met a reliability specification or that a redesigned system has improved reliability. In a demonstration test, you verify that only a certain number of failures occur in a set amount of test time.

Dialog box items
Minimum Value to be Demonstrated Scale (Weibull or expo) or location (other dists): Choose to demonstrate the minimum scale for the Weibull and exponential distributions or the minimum location for other distributions, then enter the scale or location value. Percentile: Choose to demonstrate the minimum percentile. In Percentile, enter a percentile. The percentile should be in units of time. In Percent, enter a percent associated with the percentile. The percent must be a number between 0 and 1 or a percentage between 0 and 100. Reliability: Choose to demonstrate the minimum reliability. In Reliability, enter the reliability. The reliability must be a number between 0 and 1. In Time, enter the time associated with the reliability. MTTF: Choose to demonstrate the mean time to failure (MTTF), then enter the MTTF.

8

Copyright © 2003–2005 Minitab Inc. All rights reserved.

Test Plans

Maximum number of failures allowed: Enter one or more maximum number of failures your test allows. Sample sizes: Choose to enter the number of units available for testing. Enter one or more sample sizes. Testing times for each unit: Choose to enter the amount of time available for testing. Enter one or more test durations. Note Each combination of maximum number of failures allowed and sample size or testing time will result in one test plan. You may wish to request several test plans and compare the results.

Distribution Assumptions Distribution: Choose one of seven common distributions: Weibull (default), exponential, smallest extreme value, normal, lognormal, logistic, and loglogistic. Shape (Weibull) or scale (other distributions): Enter the shape (Weibull) or scale (other distributions). For an exponential distribution, Minitab assumes a shape value of one. See Specifying planning values.

To determine testing time or sample size for a demonstration test
1 2 Choose Stat > Reliability/Survival > Demonstration Test Plans. Under Minimum Value to be Demonstrated, choose one of the following: • Scale (Weibull or expo) or location (other dists) to provide the scale of Weibull or exponential distributions or the location of other distributions, then enter the scale or location. • Percentile, then enter the percentile. In Percent, enter a number between 0 and 100 for the associated percent. • Reliability, then enter a reliability value between zero and one. In Time, enter the time. • MTTF to provide the mean time to failure, then enter the time. In Maximum number of failures allowed, enter a number greater than or equal to zero. See m-failure test plan. Under Specify values for one of the following, choose either: • Sample sizes, then enter the number of units available for testing. • Testing times for each unit, then enter each unit's test duration. Each combination of maximum number of failures allowed and sample size or testing time will result in one test plan. You may wish to request several test plans and compare the results.

3 4

Note 5 6

Under Distribution Assumptions, choose any distribution from Distribution. Then, enter an estimate of the shape or scale in Shape (Weibull) or scale (other dists). See estimating the shape or scale. If you like, use any dialog box options, then click OK.

Choosing Between a 0-Failure and an M-Failure Test
Use the table below to choose between a 0-failure and an m-failure test. A 0-failure test... Usually reduces total test time for highly reliable items. An m-failure test (m > 0)... May reduce total test time if you can run the tests sequentially. For example, if you are testing 3 units in a 1failure test and the first 2 units pass, you do not have to test the third. May not be feasible for highly reliable units. Allows you to check the assumptions of the test design. • You can estimate the shape (Weibull distribution) or scale (other distributions) and compare it to the assumed value. You can obtain a more accurate estimate of the scale (Weibull or exponential distribution) or location (other distributions).

Is more practical when failures are unlikely in a reasonable amount of time. Does not let you check the assumptions of the test design. • You cannot estimate the shape (Weibull distribution) or scale (other distributions) to compare it to the assumed value. You can estimate the scale (Weibull or exponential distribution) or location (other distributions), but your estimate may be conservative.

Does not make sense when you are likely to have at least one failure.

Has a better chance of passing than a 0-failure test when you have a marginally improved design.

M-Failure Test Plan
In an m-failure test plan, the test is successful if no more than m failures occur. For example, if m = 3, a test passes if 0, 1, 2, or 3 failures occur among N identical systems that are tested independently and have the same failure distribution. Assumptions of the m-failure test plan:

Copyright © 2003–2005 Minitab Inc. All rights reserved.

9

You can increase the power of your demonstration test in two ways: 1 Reduce your goal value. Demonstration Test Plans − Graphs Stat > Reliability/Survival > Demonstration Test Plans > Graphs Use to draw a POP (probability of passing) graph to assist you in choosing a minimum value for the parameter you want to demonstrate. it is common to have a good estimate of the shape (Weibull distribution) or scale (other distributions) parameter because this parameter is often not impacted by a redesign. 2 Type I and Type II Errors in a Demonstration Test The hypotheses for a demonstration test are: H0: The system reliability is less than or equal to a goal value. You can reduce the probability of a Type II error (β) by reducing the minimum value for the unknown parameter or by increasing the maximum number of failures your test allows. Show different test plans overlaid on the same page: Choose to display different test plans overlaid on the same page. The shape parameter is one. You can adjust the Type I error by changing the confidence level in the Options subdialog box. (Type I error) You have exceeded the goal value or a redesigned system has improved. see Choosing between a 0-failure and m-failure test. For the extreme value. For more information. Maximum X scale: Enter a value for the maximum x-axis scale. power is the probability of correctly concluding that you have demonstrated a goal value. H1: The system reliability is greater than a goal value. Reduce the minimum value you want to demonstrate. logistic. your demonstration test plan will be flawed. so that a system with high reliability has a high probability of passing the m-failure test. if your assumptions regarding this value are wrong. reducing the minimum value yields a weaker conclusion about the reliability of the systems. However. Estimating the shape or scale When running a demonstration test. but the test did not detect it. As the improvement ratio increases. If the improvement ratio is small. normal. Increasing Power The power of a test is the probability of correctly rejecting H0 when it is false. the power of the test increases. For the exponential distribution. you know the shape parameter and wish to demonstrate the scale parameter. systems that have improved or systems with high reliability values have a better chance of passing the m-failure test. See Increasing Power. (Type II error) Minitab provides testing times or sample sizes to control the Type I error (α). 10 Copyright © 2003–2005 Minitab Inc. lognormal. Show different sample sizes/testing times overlaid on the same page: Choose to display different sample sizes or testing times overlaid on the same page. and loglogistic distributions. then the goal value is too large. All rights reserved. You should consider rerunning the analysis using a range of reasonable values for the assumed parameter to see how the assumed value is impacting your conclusions. In a demonstration test. You can make either of these errors: • • The test concludes that you have exceeded the goal value. This way. Increase the maximum number of failures allowed in the m-failure test. POP Graph Use a POP (Probability of Passing) graph to choose a minimum value for the parameter you wish to demonstrate.Reliability and Survivial Analysis • • • For the Weibull distribution. Minimum X scale: Enter a value for the minimum x-axis scale. However. . but you really have not. you know the scale parameter and wish to demonstrate the location parameter. you wish to demonstrate the scale parameter. Dialog box items Probability of passing the demonstration test: Check to display a POP graph.

Choose Testing times for each unit. Select the y-axis. you are reducing the chance of making a Type II error. Note Minitab displays the likelihood of passing as a percent. the more likely the test will pass. From Distribution. By increasing power. See Increasing Power. Minitab uses the sample size and corresponding testing time to control the Type I error (α). Session window output Demonstration Test Plans Reliability Test Plan Distribution: Weibull. then enter 2000. The POP graph is a plot of the power of your test (probability of passing your test) against the improvement ratio or the improvement amount. then enter 8000. Click the Type tab and choose Probability. and choose Edit > Y Scale. In Percent. The likelihood that the test will pass depends on: • • • How much the unit's life has truly improved. You must determine the number of combustors needed to demonstrate the reliability goal using a 1-failure test plan. Demonstration Test Plans − Options Stat > Reliability/Survival > Demonstration Test Plans > Options You can enter a confidence level that Minitab will use for all confidence intervals. enter 3. Dialog box items Confidence level: Enter a number between 0 and 100. Example of creating a demonstration test plan The reliability goal for a turbine engine combustor is a 1 percentile of at least 2000 cycles. (The more the unknown true life has improved over the hypothesized value. you must edit the displayed graph. In Shape (Weibull) or scale (other dists).Test Plans The curve that appears on this graph shows you the likelihood of actually passing the demonstration test that you specified in the dialog box. choose Weibull. All rights reserved. In Maximum number of failures allowed. Target Confidence Level = 95% Failure Test 1 Testing Time 8000 Sample Size 8 Actual Confidence Level 95. You can reduce the probability of a Type II error (β) by choosing the minimum value of the unknown parameter. Testing time and sample size combinations.0.2122 Copyright © 2003–2005 Minitab Inc. The number of cycles to failure tends to follow a Weibull distribution with shape = 3. You can accumulate up to 8000 test cycles on each combustor. See Type I and Type II errors in a Demonstration Test. enter 1. To re-scale this as a probability. Choose Percentile. You can adjust the Type I error by changing the confidence level in the Options subdialog box. Shape = 3 Percentile Goal = 2000. enter 1. 1 2 3 4 5 Choose Stat > Reliability/Survival > Demonstration Test Plans. Click OK. The default is 95.) The number of failures allowed. right-click. 11 .

then enter the percent. Dialog box items Parameter to be Estimated Percentile for percent: Choose to estimate a percentile. However. then the improvement ratio = 4000/2000 = 2.5 and the probability of passing the test would increase to around 0. If the improvement ratio is greater than about two. Estimation Test Plans Estimation Test Plans Stat > Reliability/Survival > Estimation Test Plans Use to determine the number of test units that you need to estimate percentiles or reliabilities with a specified degree of precision. The data you collect can be: • • • Uncensored or complete Right-censored Interval-censored A time-censored or failure-censored test plan often gives precise results while minimizing your testing costs. then the improvement ratio would increase to 2. and the probability of passing the test would be about 0. All rights reserved. . 12 Copyright © 2003–2005 Minitab Inc. you would also be making a less powerful statement about the reliability of the turbine engine combustor. Here. The graph shows the likelihood of actually passing the test that you specified. If you reduced the value to be demonstrated to 1600.Reliability and Survivial Analysis Graph window output Interpreting the results You must test 8 combustors for 8000 cycles to demonstrate with 95.2% confidence that the first percentile is at least 2000 cycles.88. By reducing the value to be demonstrated. If the (unknown) true first percentile was 4000. • • • The probability that your 1-failure test will pass increases steadily as the improvement ratio increases from zero to two.96. the test has an almost certain chance of passing. you would increase the probability of passing the test.

Test Plans

Reliability at time: Choose to estimate the reliability at a specified time, then enter the time. Precisions as distances from bound of CI to estimate: Choose to estimate the precision between the estimate and lower bound or the estimate and upper bound, then enter the precision value. See Choosing the precision when estimating a percentile or Choosing the precision when estimating a reliability. Assumed distribution: Choose one of seven common distributions: Weibull (default), exponential, smallest extreme value, normal, lognormal, logistic, and loglogistic. Specify planning values for two of the following: Specify one value for the exponential distribution or two values for the other distributions. See Specifying Planning Values. Shape (Weibull) or scale (other distributions): Enter the shape (Weibull) or scale (other distributions). For the exponential distribution, Minitab does not expect an entry because there is no shape parameter. Scale (Weibull or expo) or location (other dists): Enter the scale (Weibull or exponential) or location (other distributions). Percentile: Enter a percentile. In Percent, enter a percent associated with the percentile. Percentile: Enter a second percentile. In Percent, enter a percent associated with the percentile.

To use an estimation test plan for estimating a percentile
1 2 3 Choose Stat > Reliability/Survival > Estimation Test Plans. Under Parameter to be Estimated, choose Percentile for percent, then enter a percent between 0 and 100. From Precisions as distances from bound of CI to estimate, choose whether you wish to provide the desired precision from the upper or lower bound to the estimate, then enter the precision. See Choosing the precision when estimating a percentile. From Distribution, choose one of the available distributions. In Specify planning values for two of the following, complete two of the following: • In Shape (Weibull) or scale (other distributions), enter the shape or scale. • In Scale (Weibull or expo) or location (other dists), enter the scale or location. • In Percentile, enter the percentile. In Percent, enter the percent. If you enter planning values for two percentiles, they must be different. If you like, use any dialog box options, then click OK.

4 5

6 Click Right Cens or Interval Cens to add any censoring information, then click OK. 7

To use an estimation test plan for estimating a reliability
1 2 3 Choose Stat > Reliability/Survival > Estimation Test Plans. Under Parameter to be Estimated, choose Reliability at time, then enter the time. From Precisions as distances from bound of CI to estimate, choose whether you wish to provide the desired precision from the upper or lower bound to the estimate, then enter the precision. See Choosing the precision when estimating a reliability. From Distribution, choose one of the available distributions. In Specify planning values for two of the following, complete two of the following: • In Shape (Weibull) or scale (other distributions), enter the shape or scale. • In Scale (Weibull or expo) or location (other dists), enter the scale or location. • In Percentile, enter the percentile. In Percent, enter the percent. If you enter planning values for two percentiles, they must be different. If you like, use any dialog box options, then click OK.

4 5

6 Click Right Cens or Interval Cens to add any censoring information, then click OK. 7

Determining sample size for estimating scale or location parameters
You may want to approximate the sample size needed to estimate the scale parameter (Weibull or exponential distribution) or the location parameter (other distributions). To do this, use an estimation test plan to obtain the sample size needed to estimate the corresponding percentile of the distribution. For example, estimating the location parameter for the normal distribution is equivalent to estimating the 50th percentile of that distribution. Use the following table to determine the percent that corresponds to the scale or location parameter for the chosen distribution. Entering this value in Percentile for percent in the Estimation Test Plan dialog box will result in the approximate sample size that you need for estimating the scale of location parameter.

Copyright © 2003–2005 Minitab Inc. All rights reserved.

13

Reliability and Survivial Analysis

Distribution Parameter to estimate Percent Normal Lognormal Logistic Loglogistic Extreme value Weibull Exponential µ exp(µ) µ exp(µ) µ θ θ 0.5 0.5 0.5 0.5 1 − e-1 1 − e-1 1 − e-1

Choosing the precision when estimating a percentile
The precision is based on the width around the confidence interval for the parameter you are estimating. The wider your confidence interval, the fewer units you need to test. For example, if you want to estimate the 10th percentile of your failure time distribution, and the lower bound is to be no more than 25 hours less than your estimate, choose Lower bound and enter 25 as your desired precision in Sample sizes or precisions as distances from bound of CI to estimate. You may want to enter a range of values for the precision, to see its impact on your sample size.

Choosing the precision when estimating a reliability
The precision is based on the width around the confidence interval for the parameter you are estimating. The wider your confidence interval, the fewer units you need to test. For example, if you want to estimate the reliability of your units at 200 hours, and the lower bound is to be a reliability that is no more than 0.025 below your estimate, choose Lower bound and enter 0.025 as your desired precision in Sample sizes or precisions as distances from bound of CI to estimate. You may want to enter a range of values for the precision, to see its impact on your sample size.

Specifying Planning Values
To create a test plan, you need information about the data you expect to collect. You can obtain planning information from: • • • • Design specifications Expert opinions Prior studies or small pilot studies Provide planning values for both unknown parameters (scale and shape or location and scale). Alternatively, you can provide planning values for one or two of the percentiles and Minitab will calculate the value of the unknown parameters. Provide a planning value for the unknown scale (Weibull or exponential distribution) or location (other distributions) parameter when the shape (Weibull distribution) or scale (other distributions) is known.

For an estimation test plan, you must do one of the following:

For an accelerated life test plan, you must provide the shape (Weibull distribution) or scale, and planning values for one of the following: • • • • Percentiles at two different stress levels One percentile and the intercept One percentile and the slope The intercept and the slope The slope represents the activation energy when the Arrhenius relationship is chosen and the assumed distribution is Weibull, exponential, lognormal, or loglogistic.

Note

Estimation Test Plans − Right Censoring
Stat > Reliability/Survival > Estimation Test Plans > Right Censoring Use the right-censoring options if your data are censored in either of these ways:

14

Copyright © 2003–2005 Minitab Inc. All rights reserved.

Test Plans
• • • • Time-censored − Test each unit for a preset amount of time. Failure-censored − Test the units until a preset proportion of failures occur. Singly-censored − All of the test units run for the same amount of time or until the same percent of units fail. Units surviving at the end of the study are considered censored data. Multiply-censored − Test units are censored at different times or in groups where a different percent of units are allowed to fail.

Your data can be either singly censored or multiply censored:

Dialog box items
Type of Censoring Time censor at: Choose for time-censored data, then enter the censoring time. If your data will be singly censored, enter one or more censoring times. If your data will be multiply censored, enter one or more columns of censoring times. Each row in a column represents a group of test units. See Time Censoring. Failure censor at percent of units failed: Choose for failure-censored data, then enter the percent of failures at which to begin censoring. If your data will be singly censored, enter one or more percents. If your data will be multiply censored, enter one or more columns of percents. Each row in a column represents a group of test units. See Failure Censoring. Allocation for Multiple Groups Equal percent per group: Choose to run the same percentage of units for each group. Percent of units run in each group: Choose to change the percentage of units run for each group, then enter the percentages.

Estimation Test Plans − Interval Censoring
Stat > Reliability/Survival > Estimation Test Plans > Interval Censoring Use interval censoring when you will be inspecting units for failures at pre-set intervals. You can space these intervals equally in time or in log time; or set intervals so that the expected number of failures in each is the same.

Dialog box items
Number of Inspections: Enter the number of inspections. Inspection times Equally spaced: Choose for equally spaced inspection times. In Last inspection time, enter the last inspection time. Equal probability: Choose for the expected proportion of failures to be the same in each interval. In Total percent of failures, enter the expected percent of failures for the entire test. Equally spaced in log time: Choose for equally spaced log inspection times. In First inspection time and Last inspection time, enter the first and last times.

Estimation Test Plans − Options
Stat > Reliability/Survival > Estimation Test Plans > Options You can assume a known shape or scale parameter. You can also enter a confidence level that Minitab will use for all confidence intervals.

Dialog box items
Assume shape (Weibull) or scale (other distributions) is known: Check if you know the shape (Weibull) or scale (other distributions) parameter. This results in a smaller sample size because Minitab assumes that you do not need to estimate this parameter. For the exponential distribution, Minitab assumes a known shape parameter of one. Confidence level: Enter the confidence level. The default is 95.0.

Example of creating an estimation test plan
You want to run a life test to estimate the 5th percentile for the life of a metal component used in a switch. You can run the test for 100,000 cycles. You expect about 5% of the units to fail by 40,000 cycles, 15% by 100,000 cycles, and the life to follow the Weibull distribution. You want the lower bound of your confidence interval to be within 20,000 cycles of your estimate. 1 2 3 Choose Stat > Reliability/Survival > Estimation Test Plans. Under Parameter to be Estimated, choose Percentile for percent, then enter 5. From Precisions as distances from bound of CI to estimate, choose Lower bound, then enter 20000.

Copyright © 2003–2005 Minitab Inc. All rights reserved.

15

Under Type of Censoring. logistic. Sample sizes or precisions as distances from bound of CI to estimate: Choose Sample size. Accelerated Life Test Plans Accelerated Life Test Plans Stat > Reliability/Survival > Accelerated Life Test Plans Use accelerated life test plans to determine the number of test units and how to allocate these units across stress levels for an accelerated life test. the default). Dialog box items Parameter to be Estimated Percentile for percent: Choose to estimate a percentile.Reliability and Survivial Analysis 4 5 From Assumed distribution. enter 5. or loge (power) transformation for the accelerating variable. Under Specify planning values for two of the following. Click Right Cens. Shape (Weibull) or scale (other distributions): Enter the shape (Weibull) or scale (other distributions). The data you collect can be: • • • Uncensored or complete Right-censored Interval-censored A time-censored or failure-censored test plan often gives precise results while minimizing testing costs. Click OK in each dialog box. lognormal. normal. See Transforming the Accelerating Variable. 15 Planning distribution: Weibull Scale = 423612. and loglogistic. then enter the percent.000 cycles of the estimate. or Upper bound. do the following: • In the first Percentile. Shape = 1. Lower bound. See Choosing the precision when estimating a percentile or Choosing the precision when estimating a reliability. 16 Copyright © 2003–2005 Minitab Inc. choose Time censor at. Arrhenius.25859 Censoring Time 100000 Precision 20000 Sample Size 74 Actual Confidence Level 95.0516 Interpreting the results To estimate the 5th percentile with a lower confidence bound within 20. In Percent. then enter the time. Minitab does not expect an entry because there is no shape parameter. then enter 100000. you must test 74 components for 100. enter 100000. Reliability at time: Choose to estimate the reliability at a specified time. For the exponential distribution. and enter either the sample size or the precision value. smallest extreme value. All rights reserved. enter 40000. Relationship: Choose linear (no transformation. Distribution: Choose one of seven common distributions: Weibull (default). In Percent. 6 7 Session window output Estimation Test Plans Type I right-censored data (Single Censoring) Estimated parameter: 5th percentile Calculated planning estimate = 40000 Target Confidence Level = 95% Planning Values Percentile values 40000. enter 15. choose Weibull. exponential. inverse temperature.000 cycles. . 100000 for percents 5. • In the second Percentile.

they must be at different stress levels. choose Reliability at time. • Lower bound. See Choosing the precision when estimating a percentile. choose one of the following: • Sample size. In Percent. In Percent. If your data are censored. complete two of the following: • In Percentile. enter the design stress. See Transforming the Accelerating Variable. choose Percentile for percent. Click Stresses. use any dialog box options. 17 . In Specify planning values for two of the following. Columns must be the same length. • Upper bound. In Test stresses. then enter the desired precision from the lower bound to the estimate. In Shape (Weibull) or scale (other distributions). • In Slope. they must be at different stress levels. See Choosing the Slope and Intercept. In Design stress. then click OK. See Choosing the precision when estimating a reliability. enter the levels of the test stresses. choose one of the available distributions. enter the stress level. enter the slope. then click OK. See Choosing the Slope and Intercept. See Choosing the Slope and Intercept. If you like. enter the design stress. In Percent. In Shape (Weibull) or scale (other distributions). 4 5 6 7 8 Copyright © 2003–2005 Minitab Inc. Slope: Enter the slope for the relationship with the accelerating variable. See Choosing the Slope and Intercept. enter the percentile. In Specify planning values for two of the following. Columns must be the same length. enter the percentile. Under Parameter to be Estimated. From Distribution. 4 5 6 7 8 9 To use an accelerated life test plan for estimating a reliability 1 2 3 Choose Stat > Reliability/Survival > Accelerated Life Test Plans. From Sample sizes or precisions as distances from bound of CI to estimate. See Choosing the precision when estimating a reliability. enter the percent. Percentile: Enter a second percentile. • Upper bound. or enter a column. • In Intercept. In Design stress.Test Plans Specify planning values for two of the following: Specify planning values for two of the model parameters. In Stress. enter a stored constant. From Relationship. From Sample sizes or precisions as distances from bound of CI to estimate. then click OK. See Choosing the Slope and Intercept. then enter the desired precision from the lower bound to the estimate. enter the slope. • In Intercept. enter the shape or scale. then enter the time. click Right Cens or Interval Cens to add censoring information. Click Stresses. choose one of the following: • Sample size. Percentile: Enter a percentile. From Relationship. enter the intercept. enter a percent associated with the percentile. choose one of the available distributions. Intercept: Enter the intercept for the relationship with the accelerating variable. enter the percent. In Stress. complete two of the following: • In Percentile. In Test stresses. • In Slope. In Stress. If your data are censored. choose one of the available relationships. then enter a percent between 0 and 100. then enter the number of units available to test. enter a stored constant. To use an accelerated life test plan for estimating a percentile 1 2 3 Choose Stat > Reliability/Survival > Accelerated Life Test Plans. See Specifying Planning Values. choose one of the available relationships. enter the shape or scale. See Choosing the Precision when estimating a percentile. enter a percent associated with the percentile. See Transforming the Accelerating Variable. From Distribution. See Choosing the Slope and Intercept. then enter the number of units available to test. enter the stress level. click Right Cens or Interval Cens to add censoring information. or enter a column. You can type the design or level of test stresses. If you choose to specify planning values for two percentiles. If you enter planning values for two percentiles. enter the intercept. enter the levels of the test stresses. then enter the desired precision from the estimate to the upper bound. If you enter planning values for two percentiles. enter the stress level. You can type the design or level of test stresses. enter the stress level. In Stress. In Percent. they must be at different stress levels. Under Parameter to be Estimated. then enter the desired precision from the estimate to the upper bound. All rights reserved. • Lower bound.

then click OK. or enter columns. however. By default. Use columns for a set of test stress levels for a series of test plans. Columns must be the same length as the columns of test stresses and must sum to 100%. See Accelerated Life Test Models. Type or enter stored constants if the percent allocations are for a single test plan and sum to 100%. or loglogistic. a common rule of thumb is that the expected number of failures at each of the test stresses should be at least four or five. It is possible. The default is 3. The slope.05. ε = random error term. In particular. . for a highly efficient test plan (one with small variance) to produce results that are not accurate. You can type the percent allocations. you can use historical estimates of the slope and intercept as planning values. All rights reserved. User Defined Allocations for each Stress Level Percent Allocations: Enter the percent of units to test at each test stress level. lognormal. See Efficiency and Accuracy of Accelerated Life Test Plans.16)] + σε Y = β0 + β1 ∗ log(accelerating variable) + σε Y = β0 + β1 ∗ accelerating variable + σε Note Choosing the Slope and Intercept If you have previously used accelerated life tests for similar experiments. 18 Copyright © 2003–2005 Minitab Inc. the results of an accelerated life test are based on obtaining enough failures at each stress level to accurately estimate the parameter of interest. Alternatively. You can type the stress levels. By default.83/° C + 273. Minitab displays three different test plans. you can provide the allocation. See Prefixed Ranges and Default Steps.01 and 0.Reliability and Survivial Analysis 9 If you like. Each column represents a separate set of test stresses. Efficiency and Accuracy of Accelerated Life Test Plans Minitab evaluates the efficiency of each plan and ranks them in order.5 to use to go through the range of each test stress. The default is 0. use any dialog box options. β0 = y-intercept (constant) β1 = regression coefficient σ = reciprocal of the shape parameter (Weibull distribution) or the scale parameter (other distributions). To obtain accurate parameter estimates. Efficiency is measured in terms of the variance of the parameter you want to estimate. Accelerated Life Test Plans − Stress Levels Stat > Reliability/Survival > Accelerated Life Test Plans > Stresses You must enter the design and test stress levels.16)] + σε Y = β0 + β1 ∗ [1/(° C + 273. β1. Test stresses: Enter one or more fixed test stress levels. is the activation energy in Arrhenius models when the assumed distribution is Weibull. Step length in search: Enter a value between 0. Number of "best" plans to output: Enter the number of test plans for Minitab to display. Dialog box items Design stress: Enter the stress level for normal use conditions. See Searching for the Optimum Proportions. Minitab will determine an "optimal" allocation of units across stress levels. enter stored constants. Type or enter stored constants if the test stress levels are for a single test plan. Accelerated Life Test Models Relationship Arrhenius Inverse temperature Loge (power) Linear where: • • • • • Y = failure time or log failure time. Use columns for a set of allocations for a series of test plans. enter stored constants. Search for the Best Allocation for each Stress Level: Check to have Minitab find the optimal allocation for each stress level. Model Y = β0 + β1 ∗ [11604. exponential. or enter columns.

Test Plans

Searching for the Optimum Proportions
The most efficient plan is only the most efficient in the specified search space. Minitab can find the most efficient or "optimum" allocation of test units in two ways: • • You specify the search space as a finite set or sets of proportions. Each column represents a different test plan. Minitab ranks those test plans according to their efficiency. Minitab searches for the optimum proportions in ranges. That is, Minitab uses a step to go from one candidate group of proportions to another. The default step length is 0.05, but you can increase or reduce the length.

Pre-fixed Ranges and Default Steps
Minitab chooses the pre-fixed ranges for the proportionate allocation of test units based on the following criteria: • • • • • • More test units are assigned to the lowest test stress. Either a large or small number of test units exists at the middle stresses. The proportionate allocation of units at a test stress is not too small relative to the others. For a two-stress design, the ranges for the proportions at the lowest and highest test stress are RL = [0.05, 0.85] and RH = [0.075, 0.5], respectively. For a three-stress design, the lowest test stress range is RL = [0.333, 0.683]. The other test stresses have a common range of R = [0.040, 0.333]. In general, if your design has K test stresses, the range for the proportions at the − lowest test stress is RL = [1/K, 1/K + 0.35] − middle stresses is R = [(1- 1/K - 0.350)/2K, 1/K] − highest test stress is R = [(1- 1/K - 0.350)/2K, 1/K] and chosen so that the complete set of proportionate allocations sums to one

The ranges change as the number of stresses change:

Accelerated Life Test Plans − Right Censoring
Stat > Reliability/Survival > Accelerated Life Test Plans > Right Censoring Use the right-censoring options if your data will be censored in either of these ways: • • Time-censored − Test each unit for a preset amount of time, which can be different for each stress level. Failure-censored − Test the units until a preset proportion of failures occurs. The proportion can be different for each stress level.

Dialog box items
Type of Censoring Time censor for each stress level: Choose for time-censored data, then enter the censoring time for each stress level, in order, from the lowest to the highest. See Time Censoring. Failure censor at percent of units failed for each stress level: Choose for failure-censored data, then enter the percent of failures at which to begin censoring for each stress level, in order, from the lowest to the highest. See Failure Censoring.

Accelerated Life Test Plans − Interval Censoring
Stat > Reliability/Survival > Accelerated Life Test Plans > Interval Censoring Use interval censoring when you will be inspecting units for failures at pre-set intervals. You can space these intervals equally in time or in log time; or set intervals so that the expected number of failures in each is the same.

Dialog box items
Number of inspections for each stress level: Enter the number of inspections for each stress level, in order, from the lowest stress level to the highest stress level. You must have the same number entries as you have test stress levels. Inspection Times Equally spaced: Choose for equally spaced inspection times. In Last inspection time for each stress level, enter the last inspection time from the lowest stress level to the highest stress level. You must have the same number entries as you have test stress levels. Equal probability: Choose for the expected proportion of failures to be the same in each interval. In Total percent of failures at each stress level, enter the expected percent of failures in the entire test from the lowest stress level to the highest stress level. You must have the same number entries as you have test stress levels.

Copyright © 2003–2005 Minitab Inc. All rights reserved.

19

Reliability and Survivial Analysis

Equally spaced in log time: Choose for equally spaced log inspection times. In First inspection time for each stress level and Last inspection time for each stress level, enter the first and last times from the lowest stress level to the highest stress level. You must have the same number entries as you have test stress levels.

Accelerated Life Test Plans − Options
Stat > Reliability/Survival > Accelerated Life Test Plans > Options You can assume a known shape or scale parameter. You can also enter a confidence level that Minitab will use for all confidence intervals.

Dialog box items
Assume shape (Weibull) or scale (other distributions) is known: Check if you know the shape (Weibull) or scale (other distributions) parameter. This results in a smaller sample size because Minitab assumes that you do not need to estimate this parameter. For an exponential distribution, Minitab assumes a known shape parameter of one. Confidence level: Enter the confidence level. The default is 95.0.

Example of creating an accelerated life test plan
You want to plan an accelerated life test to estimate the 1000-hour reliability of an incandescent light bulb at the design voltage of 110 volts. You have 20 light bulbs available to test until failure. To accelerate failures, you will run the test at 120 volts and 130 volts. You believe that a power relationship will adequately model the relationship between failure time and voltage. Historical data indicate that a lognormal distribution with a scale of 50 appropriately models light bulb failure. The planning values are 1200 for the 50th percentile at 110 volts and 600 for the 50th percentile at 120 volts. 1 Choose Stat > Reliability/Survival > Accelerated Life Test Plans. 2 3 4 5 6 Under Parameter to be Estimated, choose Reliability at time, then enter 1000. In Sample sizes or precisions as distances from bound of CI to estimate, choose Sample size, then enter 20. From Distribution, choose Lognormal. From Relationship, choose Loge (Power). In Shape (Weibull) or scale (other distributions), enter 50. Under Specify planning values for two of the following, do the following: • In the first Percentile, enter 1200. In Percent, enter 50. In Stress, enter 110. • In the second Percentile, enter 600. In Percent, enter 50. In Stress, enter 120. Click Stresses. In Design stress, enter 110. In Test stresses, enter 120 130. Click OK in each dialog box.

7 8

Session window output Accelerated Life Testing Test Plans
Uncensored data Power model Estimated parameter: Reliability at time = 1000 Calculated planning estimate = 0.501455 Design stress value = 110 Target Confidence Level = 95% Planning Values Percentile values = 1200, 600 for percents = 50, 50 at stresses = 110, 120 Planning distribution: Lognormal base e Intercept = 44.5349, Slope = -7.96617 and Scale = 50 Selected test plans: "Optimum" allocations test plans

Total available sample units = 20

20

Copyright © 2003–2005 Minitab Inc. All rights reserved.

Test Plans

1st Best "Optimum" Allocations Test Plan Test Stress 120 130 Percent Failure 100 100 Percent Alloc 65.7524 34.2476 Sample Units 13 7 Expected Failures 13 7

Standard error of the parameter of interest = 0.283150

2nd Best "Optimum" Allocations Test Plan Test Stress 120 130 Percent Failure 100 100 Percent Alloc 65 35 Sample Units 13 7 Expected Failures 13 7

Standard error of the parameter of interest = 0.283185

3rd Best "Optimum" Allocations Test Plan Test Stress 120 130 Percent Failure 100 100 Percent Alloc 70 30 Sample Units 14 6 Expected Failures 14 6

Standard error of the parameter of interest = 0.284363

Interpreting the results
To estimate the 1000-hour reliability at the design voltage of 110 volts, test 13 units until failure at 120 volts and 7 units until failure at 130 volts.

Copyright © 2003–2005 Minitab Inc. All rights reserved.

21

.

all you may know is that a part failed before a certain time (left censoring). This type of censoring is called right censoring. then use the nonparametric estimates. survival. In this case. Estimation methods Minitab provides both parametric and nonparametric methods to estimate functions. Once you have collected your data. Choosing a distribution analysis command How do you know which distribution analysis command to use? You need to consider two things: 1) the type of censoring you have. you can use the commands in this chapter to select the best distribution to use for modeling your data. You do this by estimating percentiles. and distribution parameters and by drawing survival plots. meaning their exact failure time is unknown. If no parametric distribution adequately fits your data. or within a certain interval of time (interval censoring). If a parametric distribution fits your data. Suppose you are testing how long a certain part lasts before wearing out and plan to cut off the study at a certain time. hazard. part. 23 . • • Use the parametric distribution analysis commands when you can assume your data follow a parametric distribution. and 2) whether or not you can assume a parametric distribution for your data. Use the arbitrary-censoring commands when your data are arbitrarily censored to include both exact failures and a varied censoring scheme. and interval-censoring. You can use either parametric or nonparametric estimates." or after the present time. Use the nonparametric distribution analysis commands when you cannot assume a parametric distribution. Estimation methods Estimate Parametric (assumes parametric distribution) Method Maximum likelihood Results Distribution parameters. cumulative failure plots. These methods are called parametric because you assume the data follow a parametric distribution. or organism. Parametric estimates are based on an assumed parametric distribution. For instance. survival probabilities. see Distribution Analysis Data. hazard. Censoring − Life data are often censored or incomplete in some way. available methods depend on the type of censoring. All rights reserved. or hazard plots. For parametric estimates. cumulative failure probabilities. and percentile estimates • • Right-censored parametric distribution analysis Arbitrary-censored parametric distribution analysis Copyright © 2003–2005 Minitab Inc. Minitab provides nonparametric estimates of the same functions. and then estimate the variety of functions that describe that distribution. cumulative failure. person. Your goal is to estimate the failure-time distribution of a product. then use the parametric estimates. Any parts that did not fail before the study ended are censored. survival. If you cannot find a distribution that fits your data. For details on creating worksheets for censored data. Distribution − Life data can be described using a variety of distributions. while nonparametric estimates assume no parametric distribution. you can choose either the least squares method or the maximum likelihood method. and percentile estimates Available with • • Right-censored parametric distribution analysis Arbitrary-censored parametric distribution analysis Leastsquares estimation Distribution parameters. Similarly. or how long a patient will survive after a certain type of surgery. you might want to estimate how long a part is likely to last under different conditions. For nonparametric estimates. the failure is known only to be "on the right. including right-censoring. left-censoring.Distribution Analysis Distribution Analysis Distribution Analysis Overview Use Minitab's distribution analysis commands to understand the lifetime characteristics of a product. • • Use the right-censoring commands when you have exact failures and right censored data. cumulative failure.

Interval censored How you set up your worksheet depends. . When your data have exact failures and a varied censoring scheme. For example. • Anderson−Darling statistic for the maximum likelihood and least squares estimation methods. This table describes the types of observations you can have. Type of observation Exact failure time Description You know exactly when the failure occurred. You only know that the failure The fan failed sometime between 475 occurred between two particular and 500 days. cumulative failure. or under different combinations of stress variables. including right-censoring. You only know that the failure occurred after a particular time. Goodness-of-fit statistics Minitab displays up to two goodness−of−fit statistics to help you compare the fit of distributions. Example The fan failed at exactly 500 days. Right censored The fan had not yet failed at 500 days. left-censoring. and density estimates. 24 Copyright © 2003–2005 Minitab Inc. All rights reserved.Reliability and Survivial Analysis Nonparametric (no distribution assumed) KaplanMeier Survival. cumulative failure. median residual lifetimes • Right-censored nonparametric distribution analysis Arbitrary-censored nonparametric distribution analysis Right-censored distribution overview plot Arbitrary-censored distribution overview plot • • • Turnbull Survival and cumulative failure estimates • Arbitrary-censored nonparametric distribution analysis Right-censored distribution overview plot • Distribution Analysis Data The data you gather for the distribution analysis commands are individual failure times. and hazard estimates • Right-censored nonparametric distribution analysis Right-censored distribution overview plot • Actuarial Survival. Left censored You only know that the failure The fan failed sometime before 500 occurred before a particular time. on the type of censoring you have: • • When your data consist of exact failures and right-censored observations. Suppose you are monitoring air conditioner fans to find out the percentage of fans that fail within a three-year warranty period. in part. You might also collect samples of failure times under different temperatures. Life data are often censored or incomplete in some way. see Distribution analysis (arbitrarily censored data). see Distribution analysis (right censored data). hazard. and interval-censoring. days. times. you might collect failure times for units running at a given temperature.

Here is the same data set structured both ways: Unstacked Data Drug A 20 30 43 51 57 82 85 89 Drug B 2 3 6 14 24 26 27 31 Stacked Data Drug 20 30 43 51 57 82 85 89 2 3 6 14 24 26 27 31 Group A A A A A A A A B B B B B B B B Note You cannot analyze more than one column of stacked data at a time. Arbitrarily Censored Data Distribution ID Plot Parametric distribution analysis commands You can use all parametric distribution analysis commands for both right-censored and arbitrarily-censored data. The Pearson correlation measures the strength of the linear relationship between the X and Y variables on a probability plot. Weibull. These graphs are often used before the full analysis to help choose a distribution or view summary information. logistic. you can stack all the data in one column and add a column of grouping indicators that define each sample. with higher values indicating a better fitting distribution. Command Distribution ID Plot Right Censored Arbitrarily Censored Description Draws probability plots from your choice of eleven common distributions: smallest extreme value. Distribution Overview Plot Right Censored Arbitrarily Censored Copyright © 2003–2005 Minitab Inc. normal. The commands include Parametric Distribution Analysis. Unstacked data In unstacked data. Draws a probability plot. 3-parameter lognormal. if any. so the grouping indicators must be in one column. All rights reserved. loglogistic. Like censoring indicators. These help you assess the fit of the chosen distribution and view summary graphs of your data. grouping indicators can be numbers or text. Minitab uses an adjusted Anderson−Darling statistic. each sample is in a separate column. and hazard plot in separate regions on the same graph. survival plot. because the statistic changes when a different plot point method is used. The correlation will range between 0 and 1. These plots help you determine which. lognormal. 3-parameter Weibull. Stacked vs. of the parametric distributions best fits your data. 2parameter exponential. The statistic is a weighted squared distance from the plot points to the fitted line with larger weights in the tails of the distribution. The Anderson−Darling statistic is a measure of how far the plot points fall from the fitted line in a probability plot. A smaller Anderson−Darling statistic indicates that the distribution fits the data better. probability density function. Alternatively. and 3-parameter loglogistic. and creates a Distribution ID Plot and Distribution Overview Plot.Distribution Analysis • Pearson correlation coefficient for the least squares estimation method. which performs the full analysis. 25 . exponential.

exponential. including right-.. You can enter up to 50 columns (50 different samples). Minitab estimates the functions independently for each sample. and probability plots. All the samples display on a single plot. logistic. 2-parameter exponential. Failure time Missing value symbol '∗' Time before which the failure occurred Time at start of interval during which the Time at end of interval during which the failure occurred failure occurred 26 Copyright © 2003–2005 Minitab Inc.and interval-censored data. normal.. loglogistic. 3-parameter lognormal. For systems that have more than one cause of failure. lognormal. Distribution ID Plot (Arbitrary Censoring) Stat > Reliability/Survival > Distribution Analysis (Arbitrary Censoring) > Distribution ID Plot Use Distribution ID Plot (Arbitrary Censoring) to determine which distribution best fits your data by comparing how closely the plot points lie to the best-fit lines of a probability plot. normal. For general information on life data and censoring.. see Multiple Failure Modes (Arbitrarily Censored Data). and cumulative failure probabilities.. Distribution 4: Check and choose one of eleven distributions: smallest extreme value. All the samples display on a single plot. 3-parameter Weibull. Distribution 2: Check and choose one of eleven distributions: smallest extreme value. unless you assume a common shape (Weibull) or scale (other distributions). 2-parameter exponential. Dialog box items Start variables: Enter the columns of start times. then uses that distribution to estimate percentiles. Enter your data in table form. 3-parameter Weibull. using a Start column and End column: For this observation. 3-parameter lognormal. Use all distributions: Choose to have Minitab fit all eleven distributions. Frequency columns (optional): Enter the columns of frequency data. 3-parameter Weibull. cumulative failure. or 3parameter loglogistic. or 3parameter loglogistic. logistic. then enter a column of grouping indicators. Exact failure time Right censored Left censored Interval censored Enter in the Start Column. left. Weibull (default). exponential (default). lognormal. 3-parameter lognormal. lognormal (default). Minitab analyzes systems with one cause of failure or multiple causes of failure. You can enter up to 50 samples per analysis. . 3-parameter lognormal. normal (default). your data is arbitrarily-censored. check By variable. Failure time Time that the failure occurred after Missing value symbol '∗' Enter in the End Column. Distribution 1: Check and choose one of eleven distributions: smallest extreme value.. normal. Weibull. 2-parameter exponential. exponential. hazard. Specify: Choose to fit up to four distributions. End variables: Enter the columns of end times. logistic. Distribution 3: Check and choose one of eleven distributions: smallest extreme value. loglogistic. Minitab also provides two goodness-of-fit measures to help you assess how the distribution fits your data: • • Anderson-Darling for the least squares and maximum likelihood estimation methods Pearson correlation coefficient for the least squares estimation method You can display up to 50 samples on each plot. survival probabilities. 3-parameter Weibull. or 3parameter loglogistic. or 3parameter loglogistic. By variable: If all of the samples are stacked in one column. loglogistic. exponential. All rights reserved. Weibull. lognormal. with different colors and symbols. see Distribution Analysis Data. Distribution Analysis (Arbitrarily Censored Data) When your data consist of exact failures and a varied censoring scheme. Weibull.Reliability and Survivial Analysis Parametric Distribution Analysis Right Censored Arbitrarily Censored Fits one of eleven common parametric distributions to your data. You can enter up to 50 columns (50 different samples). logistic. loglogistic. 2-parameter exponential. Also draws survival.. with different colors and symbols. which helps you compare the various functions between samples.

you can stack all the samples in one column. and so on. If all of the samples are stacked in one column. use any of the dialog box options. check By variable. enter them in Frequency columns. Interval censored between 50000 and 60000 hours. Maximum Likelihood: Choose to estimate the distribution parameters using the maximum likelihood method. If you have frequency columns. For observations with corresponding columns of frequency. Show graphs of different variables or by levels: Choose to display the graphs overlaid on the same graph or on separate graphs. For an illustration. the second start column is paired with the second end column. Estimate percentiles for these percents: Enter the additional percents for which you want to estimate percentiles. All rights reserved. 27 . 7 Distribution ID Plot (Arbitrary Censoring) − Options Stat > Reliability/Survival > Distribution Analysis (Arbitrary Censoring) > Distribution ID Plot > Options You can choose the method used to estimate the parameters. In End variables. which are estimated by fitting a regression line to the points in a probability plot. Copyright © 2003–2005 Minitab Inc. and enter a column of grouping indicators in the box. Exact failures at 30000 hours. Use: Choose to enter your own values for the X scale minimum and maximum. enter up to 50 columns of start times. see Stacked vs. which are estimated by maximizing the likelihood function. Dialog box items Estimation Method Least Squares (failure time(X) on rank(Y)): Choose to estimate the distribution parameters using the least squares (XY) method. If you like.Distribution Analysis This data set illustrates tabled data. The first start column is paired with the first end column. Left censored at 10000 hours. You can enter individual percents (0 < P < 100) or a column of percents. You can also estimate percentiles for specified percents. Use default values: Choose to use the default values for the minimum and maximum X scale. then set up a column of grouping indicators. Alternatively. you can use separate columns for each sample. see Using frequency columns. In Start variables. Start * 10000 20000 30000 30000 40000 50000 50000 60000 70000 80000 90000 End 10000 20000 30000 30000 40000 50000 50000 60000 70000 80000 90000 * Right censored at 90000 hours. Unstacked data. and add your own title. enter up to 50 column of end times. which can be numbers or text. Do one of the following: • Choose Use all distributions to create probability plots for all eleven distributions. specify the x-axis minimum and maximum. When you have more than one sample. then click OK. To make a distribution ID plot (arbitrarily censored data) 1 2 3 4 5 6 Choose Stat > Reliability/Survival > Distribution Analysis (Arbitrary Censoring) > Distribution ID Plot. Minimum X scale: Enter a value for the minimum X scale. • Choose Specify to create up to four probability plots with the distributions of your choice.

015 528.1 27495. Example of a Distribution ID Plot for arbitrarily-censored data Suppose you work for a company that manufactures tires.07 1335.2 5220.3 469.960 6.000 miles) to see if the tire has failed.08 1049.3 497.7 36103. All rights reserved. You inspect each good tire at regular intervals (every 10.5 17018.1 2541.7 28 Copyright © 2003–2005 Minitab Inc.38 33699." or wear down to 2/32 of an inch of tread.70 4395. enter Freq.43 25368.MTW. You are especially interested in knowing how many of the tires last past 45.1 35912.3 75764.948 0.325 Correlation Coefficient 0.18 37033.06 15.000 miles. Title: To replace the default title with your own title. Session window output Distribution ID Plot: Start = Start and End = End Using frequencies in Freq Goodness-of-Fit Anderson-Darling (adj) 2. In Start variables.68 30366. In End variables.28 47099.466 1356. Distribution ID Plot − Arbitrary Censoring can help you choose that distribution. Leave the first three distributions at the default. Choose Specify.8 71301.8 4916.6 2698.81 41780.0 64458.411 2.998 Distribution Weibull Lognormal Exponential Smallest Extreme Value Table of Percentiles Standard Error 4005. enter End. . 1 2 3 4 5 Open the worksheet TIREWEAR.3 36461.6336 1700. which requires you to specify the distribution for your data.1 19317.4 66931.5 34342.387 2.1 5542.16 27623.82 77. choose Smallest extreme value.8 Distribution Weibull Lognormal Exponential Smallest Extreme Value Weibull Lognormal Exponential Smallest Extreme Value Weibull Lognormal Exponential Smallest Extreme Value Weibull Lognormal Exponential Smallest Extreme Value Percent 1 1 1 1 5 5 5 5 10 10 10 10 50 50 50 50 Percentile 15065.6 72547.1 20222.67 4050.1 25001. enter Start.0 44982.0 32346.1 62077.880 * 0.24 14299.5 28977.678 4461.4 69867.2114 2509. In Frequency columns.6 37732.02 1238. Click OK. Choose Stat > Reliability/Survival > Distribution Analysis (Arbitrary Censoring) > Distribution ID Plot.9 30670.0 21927.60 1249.6 73793.954 9380.8 33188.8 72756. You plan to get this information by using Parametric Distribution Analysis (Arbitrary Censoring).757 95% Normal CI Lower Upper 8946.3 2393. type the desired text in this box. You are interested in finding out how many miles it takes for various proportions of the tires to "fail. From Distribution 4. then enter the data into the Minitab worksheet.9 30237.70 159.88 1504.09 635.Reliability and Survivial Analysis Maximum X scale: Enter a value for the maximum X scale.19 44439.50 1333.

Distribution Analysis Table of MTTF Standard Error 2589. and 95% confidence intervals Table of MTTFs (mean time to failures). Output − Distribution ID Plot The default output consists of: • • • • Goodness-of-fit statistics for the chosen distributions Table of percents and their percentiles.6 52603. Here.0 68059.3 76858. A smaller Anderson-Darling statistic means that the distribution provides a better fit.48 1513. standard errors. The table of percentiles and MTTFs allow you to see how your conclusions may change with different distributions.36 1571.6 70748.0 69403. You can also compare the Anderson-Darling goodness-of-fit values to determine which distribution best fits the data. their standard errors. All rights reserved.7 70696. and 95% confidence intervals Probability plots for the chosen distributions Copyright © 2003–2005 Minitab Inc. 29 .3 Distribution Weibull Lognormal Exponential Smallest Extreme Value Mean 74139. so the extreme value distribution would be a good choice when running the parametric distribution analysis.90 95% Normal CI Lower Upper 69234.9 73712.1 46666. the Anderson-darling values for the smallest extreme value distribution are lower than the Anderson-Darling values for other distributions.9 Distribution ID Plot for Start Graph window output Interpreting the results The points fall approximately on the straight line on the smallest extreme value probability plot.52 685.6 79392.8 49546. thus supporting your conclusion that the smallest extreme value distribution provides the best fit.

normal.7339 21.48126 23.542291 1.5305 23.5140 2.0893 28.68499 27. The commands include Parametric Distribution Analysis.85714 2.7777 56.589 Correlation Coefficient 0.21312 3.63143 95% Normal CI Lower Upper 13.11939 34.2625 7. if any.9267 Table of MTTF Standard Error 2. which performs the full analysis.6233 3.0842 Distribution Weibull Lognormal Exponential Normal Weibull Lognormal Exponential Normal Weibull Lognormal Exponential Normal Weibull Lognormal Exponential Normal Percent 1 1 1 1 5 5 5 5 10 10 10 10 50 50 50 50 Percentile 17. 2parameter exponential. 3-parameter Weibull.9267 Distribution ID Plot for Temp80 Overview Plot Parametric distribution analysis commands You can use all parametric distribution analysis commands for both right-censored and arbitrarily-censored data.6314 95% Normal CI Lower Upper 51.76766 20.4812 28.97051 2. Distribution ID Plot: Temp80 Goodness-of-Fit Anderson-Darling (adj) 67.44778 2.2638 27.1273 27.4005 51.00309 6.8299 33.Reliability and Survivial Analysis For example. and 3-parameter loglogistic.519 67. Command Distribution ID Plot Right Censored Arbitrarily Censored Description Draws probability plots from your choice of eleven common distributions: smallest extreme value.90747 3.3846 56. These plots help you determine which.4791 69.5539 69.73184 33.2497 54.3722 33.2348 37. .9000 56.6552 40. 3-parameter lognormal.5948 0. exponential.41131 2.6179 61.7720 63.5033 50.68093 1.7692 62.7733 22.4955 62.5452 73.7394 3.5365 62.986 0.1830 38. lognormal.3558 50. and creates a Distribution ID Plot and Distribution Overview Plot.987 Distribution Weibull Lognormal Exponential Normal Table of Percentiles Standard Error 2.0123 0.2361 5.0842 Distribution Weibull Lognormal Exponential Normal Mean 56.46626 0.38602 2.8663 56.1804 62.98090 2.590591 3.71588 2. Weibull. of the parametric distributions best fits your data. loglogistic. 30 Copyright © 2003–2005 Minitab Inc.982 * 0.7692 33. logistic.5156 40.2431 17.76414 27.61498 0.7571 28. All rights reserved.9575 99.737540 15.5279 53. These graphs are often used before the full analysis to help choose a distribution or view summary information.8295 11.8063 51.656 71.9548 10.115720 4.606 67.2635 51.6859 23.9307 2.38997 7.9618 34.4020 5.

exponential. and if you have arbitrarily-censored data. cumulative failure. Minitab displays a Turnbull survival plot or an Actuarial survival plot and hazard plot. see Distribution Analysis Data. a probability density function. unless you assume a common shape (Weibull) or scale (other distributions). The parametric display includes a probability plot (for a selected distribution). see one of the Distribution Analysis Commands.and interval-censored data. By variable: If all of the samples are stacked in one column. and a hazard plot. or a nonparametric overview plot. check By variable. Distribution Overview Plot (Arbitrary Censoring) Stat > Reliability/Survival > Distribution Analysis (Arbitrary Censoring) > Distribution Overview Plot Use Distribution Overview Plot to generate a layout of plots that allow you to view your life data in different ways on one graph. Weibull (default). using a Start column and End column: For this observation. with different colors and symbols. 2-parameter exponential. All rights reserved. Failure time Missing value symbol '∗' Time before which the failure occurred Time at start of interval during which the Time at end of interval during which the failure occurred failure occurred Copyright © 2003–2005 Minitab Inc. or 3-parameter loglogistic. and hazard plot in separate regions on the same graph. and probability plots. logistic. and cumulative failure probabilities. Failure time Time that the failure occurred after Missing value symbol '∗' Enter in the End Column. 31 . Parametric analysis: Choose to perform a parametric distribution analysis. in different colors and symbols. These help you assess the fit of the chosen distribution and view summary graphs of your data. All the samples display on a single plot. These functions are all typical ways of describing the distribution of failure time data. 3-parameter lognormal. which helps you compare their various functions. survival plot. You can draw a parametric overview plot by selecting a distribution for your data. Frequency columns (optional): Enter the columns of frequency data. left.. Dialog box items Start variables: Enter the columns of start times. For systems that have more than one cause of failure. then uses that distribution to estimate percentiles. 3-parameter Weibull.Distribution Analysis Distribution Overview Plot Right Censored Arbitrarily Censored Parametric Distribution Analysis Right Censored Arbitrarily Censored Draws a probability plot. and enter a column of grouping indicators in the box. End variables: Enter the columns of end times. For general information on life data and censoring.. The nonparametric display depends on the type of data: if you have right-censored data Minitab displays a Kaplan-Meier survival plot and a hazard plot or an Actuarial survival plot and hazard plot.. Distribution: Choose one of eleven distributions: smallest extreme value. Distribution Analysis (Arbitrarily Censored Data) When your data consist of exact failures and a varied censoring scheme. Also draws survival.. loglogistic. Enter your data in table form. To draw these plots with more information. Minitab estimates the functions independently for each sample. You can enter up to fifty samples per analysis. Fits one of eleven common parametric distributions to your data. including right-... see Multiple Failure Modes (Arbitrarily Censored Data). lognormal. You can enter up to 50 columns (50 different samples). normal. which helps you compare the various functions between samples. You can enter up to 50 columns (50 different samples). Nonparametric analysis: Choose to perform a nonparametric distribution analysis. All of the samples display on a single plot. Minitab analyzes systems with one cause of failure or multiple causes of failure. a survival (or reliability) plot. survival probabilities. Exact failure time Right censored Left censored Interval censored Enter in the Start Column. You can enter up to 50 samples per analysis. Minitab estimates the functions independently for each sample. your data is arbitrarily-censored. hazard. probability density function.

enter the columns in Frequency columns. All rights reserved.Reliability and Survivial Analysis This data set illustrates tabled data. If you have frequency columns. In Start variables. . Start * 10000 20000 30000 30000 40000 50000 50000 60000 70000 80000 90000 End 10000 20000 30000 30000 40000 50000 50000 60000 70000 80000 90000 * Right censored at 90000 hours. If all of the samples are stacked in one column. which can be numbers or text. use any of the dialog box items. Exact failures at 30000 hours. To make a distribution overview plot (arbitrarily-censored data) 1 2 3 4 5 6 Choose Stat > Reliability/Survival > Distribution Analysis (Arbitrary Censoring) > Distribution Overview Plot. choose to plot one of eleven distributions. For an illustration. then click OK. For observations with corresponding columns of frequency. When you have more than one sample. which are estimated by maximizing the likelihood function. see Stacked vs. Choose to draw a parametric or nonparametric plot: • Parametric plot− Choose Parametric analysis. • Nonparametric plot− Choose Nonparametric analysis. enter up to 50 columns of start times. If you like. You can also specify the x-axis minimum and maximum. Alternatively. Title: To replace the default title with your own title. see Using frequency columns. you can stack all the samples in one column. type the desired text in this box. Maximum Likelihood: Choose to estimate the distribution parameters using the maximum likelihood method. Left censored at 10000 hours. 7 Distribution Overview Plot (Arbitrary Censoring) − Options Stat > Reliability/Survival > Distribution Analysis (Arbitrary Censoring) > Distribution Overview Plot > Options You can choose the method used to estimate the parameters. Use: Choose to enter your own values for the X scale minimum and maximum. Minimum X scale: Enter a value for the minimum X scale. Dialog box items Estimation Method Least Squares (failure time(X) on rank(Y)): Choose to estimate the distribution parameters using the least squares (XY) method. From Distribution. Maximum X scale: Enter a value for the maximum X scale. which are estimated by fitting a regression line to the points in a probability plot. and enter a column of grouping indicators in the box. Unstacked data. check By variable. Minimum and Maximum X Scale: Use default values: Choose to use the default values for the minimum and maximum X scale. and add your own title. Show graphs of different variables or by levels Choose to display the graphs overlaid on the same graph or on separate graphs. Interval censored between 50000 and 60000 hours. enter up to 50 columns of end times. then set up a column of grouping indicators. 32 Copyright © 2003–2005 Minitab Inc. you can use separate columns for each sample. In End variables.

33 . enter Start. Click OK. Session window output Distribution Overview Plot: Start = Start and End = End Using frequencies in Freq Goodness-of-Fit Anderson-Darling (adj) 2. In Distribution." or wear down to 2/32 of an inch of tread.MTW.000 miles. Choose Stat > Reliability/Survival > Distribution Analysis (Arbitrary Censoring) > Distribution Overview Plot. You plan to get this information by using Parametric Distribution Analysis (Arbitrary Censoring). enter Freq.325 Correlation Coefficient 0. In Start variables. You are especially interested in knowing how many of the tires last past 45. but first you want to have a quick look at your data from different perspectives. enter End. All rights reserved. then enter the data into the Minitab worksheet. In End variables.Distribution Analysis Example of a Distribution Overview Plot with arbitrarily-censored data Suppose you work for a company that manufactures tires.000 miles) to see if the tire has failed. choose Smallest extreme value. 1 2 3 4 5 Open the worksheet TIREWEAR. You are interested in finding out how many miles it takes for various proportions of the tires to "fail. You inspect each good tire at regular intervals (every 10. In Frequency columns.998 Distribution Smallest Extreme Value Distribution Overview Plot for Start Graph window output Copyright © 2003–2005 Minitab Inc.

Reliability and Survivial Analysis Interpreting the results These four plots describe the failure rate for tires over time.Actuarial survival plot . Distribution Overview Plot: Temp80 Goodness-of-Fit Anderson-Darling (adj) 67. Parametric hazard plot. if any. you can determine that approximately 90% of the tires last past 45. which displays estimates of the cumulative distribution function F(y) vs. Weibull. you get: • • • The Kaplan-Meier survival estimates. which displays the hazard function or instantaneous failure rate. Parametric survival and hazard estimates are based on a fitted distribution and the curve will therefore be smooth. you get: • • • • • • Goodness-of-fit statistics for the chosen distribution Probability plot. When you select a parametric display. f(y)/(1-F(y)) vs.606 Correlation Coefficient 0. failure time. Turnbull survival estimates.Nonparametric hazard plot based on the empirical hazard function For right-censored data with Actuarial method . These plots help you determine which. For right-censored data with Kaplan-Meier method . 3-parameter lognormal. lognormal. or f(y). Parametric survival (or reliability) plot. and 3-parameter loglogistic. Probability density function. With these plots.Nonparametric hazard plot based on the empirical hazard function When you select a nonparametric display. All rights reserved. of the parametric distributions best fits your data. The commands include Parametric Distribution Analysis. For example.Actuarial survival plot . exponential. which displays the survival (or reliability) function 1-F(y) vs. which displays the curve that describes the distribution of your data. Command Distribution ID Plot Right Censored Arbitrarily Censored Description Draws probability plots from your choice of eleven common distributions: smallest extreme value.Nonparametric hazard plot based on the empirical hazard function For arbitrarily-censored data with Turnbull method . failure time. 2parameter exponential. loglogistic. logistic. . normal.000 miles. and creates a Distribution ID Plot and Distribution Overview Plot.Turnbull survival plot For arbitrarily-censored data with Actuarial method . failure time. and empirical hazard function change values only at exact failure times. These graphs are often used before the full analysis to help choose a distribution or view summary information.Kaplan-Meier survival plot .986 Distribution Weibull Distribution Overview Plot for Temp80 Parametric Distribution Analysis Parametric distribution analysis commands You can use all parametric distribution analysis commands for both right-censored and arbitrarily-censored data. which performs the full analysis. 34 Copyright © 2003–2005 Minitab Inc. Output − Distribution Overview Plot The distribution overview plot display differs depending on whether you select the parametric or nonparametric display. so the nonparametric survival and hazard curves are step functions. 3-parameter Weibull.

evaluate the appropriateness of the distribution. and hazard plot in separate regions on the same graph. which helps you compare the various functions between samples.Distribution Analysis Distribution Overview Plot Right Censored Arbitrarily Censored Parametric Distribution Analysis Right Censored Arbitrarily Censored Draws a probability plot. survival probabilities. Minitab analyzes systems with one cause of failure or multiple causes of failure. then uses that distribution to estimate percentiles.. probability density function. hazard. exponential. survival probabilities. Frequency columns (optional): Enter the columns of frequency data. These help you assess the fit of the chosen distribution and view summary graphs of your data.. and enter a column of grouping indicators in the box Assumed distribution: Choose one of eleven common distributions: smallest extreme value. Also draws survival. normal. and probability plots.. cumulative failure. lognormal 3-parameter lognormal.. Dialog box items Start variables: Enter the columns of start times. estimate percentiles. To compare the fits of different distributions. your data is arbitrarily-censored. using a Start column and End column: For this observation.. and cumulative failure probabilities. Enter your data in table form. Exact failure time Right censored Left censored Interval censored Enter in the Start Column. see Distribution ID Plot (Arbitrary Censoring). see Multiple Failure Modes (Arbitrarily Censored Data). survival plot. with different colors and symbols. All the samples display on a single plot. You can enter up to 50 columns (50 different samples). You can fit one of eleven common distributions to your data. logistic. If no parametric distribution fits your data. cumulative failure. 2-parameter exponential. left. which draws four probability plots in separate regions on the same graph. For systems that have more than one cause of failure. check By variable. Minitab estimates the functions independently for each sample. see Distribution Analysis Data. and draw survival. and 3 parameter loglogistic distributions.. To view your data in different ways. Use the probability plot to see if the distribution fits your data. 35 . 3parameter Weibull. Failure time Missing value symbol '∗' Time before which the failure occurred Time at start of interval during which the Time at end of interval during which the failure occurred failure occurred This data set illustrates tabled data. see Distribution Overview Plot (Arbitrary Censoring). Weibull (default). unless you assume a common shape (Weibull) or scale (other distributions). Distribution Analysis (Arbitrarily Censored Data) When your data consist of exact failures and a varied censoring scheme. and cumulative failure probabilities. Failure time Time that the failure occurred after Missing value symbol '∗' Enter in the End Column. By variable: If all of the samples are stacked in one column. hazard. use one of the nonparametric distribution analysis commands. For general information on life data and censoring. All rights reserved. Parametric Distribution Analysis (Arbitrary Censoring) Stat > Reliability/Survival > Distribution Analysis (Arbitrary Censoring) > Parametric Distribution Analysis Use Parametric Distribution Analysis-Arbitrary Censoring when you have data that is arbitrarily censored or actual failure times. You can enter up to 50 samples per analysis.and interval-censored data. For observations with corresponding columns of frequency. loglogistic. and probability plots. You can enter up to 50 columns (50 different samples). End variables: Enter the columns of end times. Fits one of eleven common parametric distributions to your data. Copyright © 2003–2005 Minitab Inc. see Using frequency columns. including right-.

Left censored at 10000 hours. such as hours.Reliability and Survivial Analysis Start * 10000 20000 30000 30000 40000 50000 50000 60000 70000 80000 90000 End 10000 20000 30000 30000 40000 50000 50000 60000 70000 80000 90000 * Right censored at 90000 hours. check By variable. In Start variables. Choosing a Distribution with a Threshold Parameter The threshold parameter. the distribution starts at the origin. A negative γ indicates that failures have occurred prior to the beginning of a test. which can be numbers or text. enter up to 50 columns of start times. Unstacked data. Interval censored between 50000 and 60000 hours. the distribution starts to the right of the origin. Exact failures at 30000 hours. provides an estimate of the earliest time a failure may occur. The threshold parameter locates the distribution along the time scale and has the same units as time. If all of the samples are stacked in one column. 2-parameter exponential. Choose a distribution with a threshold parameter (3-parameter Weibull. All rights reserved. enter them in Frequency columns. you can stack all the samples in one column. When γ > 0. 3-parameter lognormal. see Stacked vs. • • • When γ = 0. In End variables. or 3-parameter loglogistic) when you want to estimate the earliest time-to-failure. γ . enter up to 50 columns of end times. use any of the dialog box options. and enter a column of grouping indicators in the box. The two probability plots show the same 36 Copyright © 2003–2005 Minitab Inc. . then click OK. The period from 0 to γ is the failure free operating period. When γ < 0. If you have frequency columns. For an illustration. the distribution starts to the left of the origin. Alternatively. When you have more than one sample. To do a parametric distribution analysis (arbitrarily censored data) 1 2 3 4 5 6 Choose Stat > Reliability/Survival > Distribution Analysis (Arbitrary Censoring) > Parametric Distribution Analysis. If you like. or cycles. miles. then set up a column of grouping indicators. you can use separate columns for each sample.

The values in the Percentile column are estimates of the times at which the corresponding percent of the units failed. 80. Percentiles By what time do half of the engine windings fail? How long until 10% of the blenders stop working? You are looking for percentiles. 37 . In the Estimate subdialog box.2590 18.2270 As shown in the first row of the table. 30.86263 17. we requested a survival probability for engine windings running at 70 months: Copyright © 2003–2005 Minitab Inc. we entered failure times (in months) for engine windings.48898 3. Survival Probabilities − Parametric Distribution Analysis What is the probability of an engine winding running past a given time? How likely is it that a cancer patient will live five years after receiving a certain drug? You are looking for survival probabilities.78453 3. which are estimates of the proportion of units that survive past a given time.23157 3.7601 12. When you request survival probabilities in the Estimate subdialog box.0765 13. In this example. Minitab displays the percentiles 1-10.6193 16.6767 24. at about 10 months (Percentile). 1% of the windings failed. 50. Here. The table also includes standard errors and approximate 95.4489 Standard Error 2.0% Normal CI Lower Upper 5. Note The threshold parameter is assumed fixed when calculating confidence intervals with the 3-parameter lognormal and 2-parameter exponential distributions. for example. 60. The Weibull does not account for the threshold parameter and displays as a curve on probability paper. The parametric distribution analysis commands automatically display a table of percentiles in the Session window. By default.5009 27. 20. 70. The 3-parameter Weibull adjusts for γ and the points appear straighter.6834 10. You can also request percentiles to be added the default table. and 90-99. Table of Percentiles Percent 1 2 3 4 Percentile 10. the parametric distribution analysis commands display them in the Session window.55426 21.66352 95.Distribution Analysis data fit to a Weibull and a 3-parameter Weibull distribution. All rights reserved. 40.3193 8.0% confidence intervals for each percentile. you can specify a different confidence level for all confidence intervals.

This option is available for both right-censored and arbitrarily censored data. Dialog box items Estimation Method Least Squares (failure time(X) on rank(Y)): Choose to estimate the distribution parameters using the least squares (XY) method. or enter a list of values for the shape or scale that is equal in length to the number of variables.0%.0% Normal CI Time Probability Lower 70. which maximizes the likelihood function. For more information on selecting estimation methods.0000 0. Parametric Distribution Analysis − Estimate Stat > Reliability/Survival > Distribution Analysis (Right Censoring) or Distribution Analysis (Arbitrary Censoring) > Parametric Distribution Analysis > Estimate Choose the estimation used to estimate percentiles and survival probabilities for specified values.5222 As shown in the table above. 38 Copyright © 2003–2005 Minitab Inc. Failure Mode Options Use all failure modes: Choose to include all failure modes in the analysis. then choose the corresponding distribution for each failure mode. use an asterisk (*) for numeric columns or a space for text columns. Estimate percentiles for these additional percents: Enter the additional percents for which you want to estimate percentiles. which fits a regression line to the points in a probability plot. Midpoint of intervals: Choose if a failure for any one failure mode causes the experiment to end. 40. Maximum Likelihood: Choose to estimate the distribution parameters using the maximum likelihood method. The default is 95. Bayes analysis Set shape (slope−Weibull) or scale (1/slope−other dists) at: Enter a value for the shape or scale parameter for all the response variables. To represent censored observations in the failure mode column. You can enter individual percents (0 < P < 100) or a column of percents. Estimate cumulative failure probabilities: Choose to estimate cumulative failure probabilities. . Minitab estimates threshold parameters if you do not provide values to use. Right endpoint of intervals: Choose if the experiment continues until the right endpoint when a failure in the interval occurs. Change Distribution for Levels Use only if distribution is different from that selected in the main dialog box. Create right censored observations using (Only available with arbitrary censoring. Set threshold at: Enter a value for the threshold parameter for all the variables.4076 0. Estimate survival probabilities: Choose to estimate survival probabilities. Assume common shape (slope−Weibull) or scale (1/slope−other dists): Check to estimate the parameters while assuming a common shape or scale parameter. Confidence level: Enter a confidence level for all of the confidence intervals.2894 Upper 0. Estimate probabilities for these times (values): Enter one or more times or a column of times for which you want to calculate survival probabilities or cumulative failure probabilities.76% of the engine windings last past 70 months.Reliability and Survivial Analysis Table of Survival Probabilities 95.) Use to determine how Minitab will create right-censored observations for other failure modes when data are interval censored. Dialog box items Use failure mode columns: Enter the columns containing the failure modes. or enter a list of values equal to the number of response variables. All rights reserved. see Least squares estimates versus maximum likelihood estimates. Eliminate failure modes: Enter the failure modes to exclude from the analysis. Parametric Distribution Analysis − Failure Mode Stat > Reliability/Survival > Distribution Analysis (Right Censoring) or Distribution Analysis (Arbitrary Censoring) > Parametric Distribution Analysis > FMode Use to estimate the overall reliability of your system when multiple causes of failure exist in order to investigate the reliability of the individual failure modes. Level: Enter failure mode. Use failure modes: Enter the failure modes to include in the analysis. You can also enter a confidence level that Minitab will use for all confidence intervals.

To request parametric survival probabilities 1 2 3 In the main dialog box. To choose the method for estimating parameters 1 2 3 In the main dialog box. the maximum likelihood parameter estimates may exist for a Weibull distribution. click Estimate. In Estimate probabilities for these times (values). Click OK. 3 For more information. All rights reserved. Under Estimation Method. for each set of distribution parameters. Least squares estimates versus maximum likelihood estimates Least squares estimates are calculated by fitting a regression line to the points in a probability plot. In Confidence level. MLE allows you to perform an analysis when there are no failures. When there is only one failure and some rightcensored observations. enter a value. LSXY is more accurate than MLE. To draw conclusions when you have few or no failures 1 3 2 In the main dialog box. To request additional percentiles 1 2 3 In the main dialog box. 39 . click Estimate. click Estimate. the chance that the true distribution has the parameters based on the sample. You can enter individual percents (0 < P < 100) or a column of percents. choose Maximum Likelihood. The line is formed by regressing time to failure or log (time to failure) (X) on the transformed percent (Y). enter the threshold value. click Estimate.Distribution Analysis Confidence intervals: Choose to use two-sided confidence intervals (the default) or just an upper or lower confidence interval. Click OK. MLE tends to overestimate the shape parameter for a Weibull distribution and underestimate the scale parameter in other distributions. The likelihood function describes. To change confidence levels 1 2 3 In the main dialog box. Click OK. under Bayes Analysis: • In Set shape (slope−Weibull) or scale (1/slope−other dists) at. Therefore. Maximum likelihood estimates are calculated by maximizing the likelihood function. see Drawing conclusions when you have few or no failures. The maximum likelihood estimation method has attractive mathematical qualities. Maximum likelihood (MLE) • • • Distribution parameter estimates are more precise than least squares (XY). enter one or more times. click Estimate. Click OK. enter the additional percents for which you want to estimate percentiles. Click OK. Copyright © 2003–2005 Minitab Inc. Under Estimation Method. Depending on your distribution. For small or heavily censored sample. enter the shape or scale value. Here are the major advantages of each method: Least squares (LSXY) • • Better graphical display to the probability plot because the line is fitted to the points on a probability plot. In Estimate percentiles for these additional percents. choose Least Squares (the default) or Maximum Likelihood. • In Set threshold at. or a column of times for which you want to calculate survival probabilities. MLE will tend to overestimate the low percentiles.

survival probabilities. Estimate at least one parameter by: − Fixing the shape OR scale − Fixing the threshold − Fixing the shape OR scale AND threshold Drawing conclusions when you have few or no failures When you have few or no failures. scale. Least squares estimates are calculated by fitting a regression line to the points in a probability plot. if your values are an appropriate choice. scale. the algorithm terminates. you can use historical values for distribution parameters to improve your analysis. you can fix all parameter or you can estimate at least one parameter. . you conclude that your product meets specifications and terminate the test. − For 2-parameter exponential. When you set historical parameters. scale. your reliability specifications require that the 5th percentile is at least 12 months. if you like. enter scale and threshold. no estimation is done. The data are right-censored. and cumulative failure probabilities. If the maximum number of iterations is reached before convergence. and threshold. If you collect life data and have no failures. − For 3-parameter Weibull. Otherwise. If the lower confidence bound for the 5th percentile is 13. The maximum likelihood solution may not converge if the starting estimates are not in the neighborhood of the true solution. Estimating the distribution parameters You can choose to estimate the parameters using either the least squares (XY) method or the maximum likelihood method (modified Newton-Raphson algorithm). If your data are from an exponential distribution. In this case. and threshold. If your data come from a three−parameter Weibull or two−parameter exponential.0% confidence intervals for the following hypothesis tests: 40 Copyright © 2003–2005 Minitab Inc. enter location and scale. Minitab performs Wald Tests [7] and provides Bonferroni 95. then there is more support for your conclusions. percentiles. you must also provide a historical value for the threshold parameter. Providing historical parameters makes the resulting analysis more precise. − For other 3-parameter distributions. and then examine the lower confidence bound to substantiate that the product is at least as good as specifications require. − For exponential. enter mean. or threshold parameters. If your data come from a Weibull or exponential distribution. The maximum likelihood method will be used to estimate parameters. Minitab automatically assigns a shape parameter of 1. you may want to use the more conservative estimates or consider the advantages of both approaches and make a choice for your problem. all results − such as the percentiles − are based on the parameters you enter. location.1 months. See Demonstration Test Plans to determine the optimal testing time or number of test units to use. and whether two or more samples share the same shape. The line is formed by regressing time to failure or log (time to failure) (X) on the transformed percent (Y). Or. you can: • Enter starting values for the algorithm. You provide a historical value for the shape parameter (Weibull). enter location. Minitab obtains maximum likelihood estimates through an iterative process. You run a Bayes analysis on data with no failures. Parametric Distribution Analysis − Test Stat > Reliability/Survival > Distribution Analysis (Right Censoring) or Distribution Analysis (Arbitrary Censoring) > Parametric Distribution Analysis > Test Tests whether distribution parameters for a sample equal specified values. Note For example. you can set your own parameters. When Minitab estimates the parameters using the maximum likelihood method. • Change the maximum number of iterations for reaching convergence (the default is 20).Reliability and Survivial Analysis When possible. Minitab can still analyze when all of the following are met: • • • • The data come from a Weibull or exponential distribution. if the results are consistent. both methods should be tried. you can do a Bayes analysis to obtain lower confidence bounds for parameters. All rights reserved. − For other 2-parameter distributions. enter shape and scale. You may want to enter reasonable starting values for parameter estimates: − For Weibull. enter shape.

Test for equal threshold: Check to test whether two or more samples have the same threshold. enter the parameter of a historical distribution. click Test. enter the test value. Test for equal scale (Weibull or expo) or location (other distributions): Check to test whether two or more samples have the same scale or location. • In Test threshold equal to. Do one or more of the following: • In Test shape (slope−Weibull) or scale (1/slope−other dists) equal to. For 3-parameter distributions. or threshold parameters Dialog box items Consistency of Sample with Value Test shape (slope−Weibull) or scale (1/slope−other dists) equal to: Enter a test value to compare the sample's shape (Weibull) or scale (other distributions). • Check Test for equal threshold. Do one or more of the following: • Check Test for equal shape (slope−Weibull) or scale (1/slope−other distributions). click Test. • Check Test for equal scale (Weibull or expo) or location (other distributions). Do the following: • In Test shape (slope−Weibull) or scale (1/slope−other dists) equal to. or threshold) are consistent with specified values Whether the sample comes from the historical distribution Whether two or more samples come from the same population Whether two or more samples share the same shape. to test the same. Do the following: • Check Test for equal shape (slope−Weibull) or scale (1/slope−other dists). Equality of Parameters Test for equal shape (slope−Weibull) or scale (1/slope−other distributions): Check to test whether two or more samples have the same shape or scale. Note For 2-parameter distributions. Test threshold equal to: Enter a test value to compare the sample's threshold. • Check Test for equal scale (Weibull or expo) or location (other distributions). click Test. click Test. Click OK.Distribution Analysis • • • • Whether the distribution parameters (scale. enter the test value. Click OK. • In Test scale (Weibull or expo) or location (other dists) equal to. location. location. enter the test value. check the first two Equality of Parameters options to test whether two or more samples come from the same population. 41 . shape. • Check Test for equal threshold. 3 To determine whether two or more samples come from the same population 1 2 In the main dialog box. To compare distribution parameters to a specified value 1 2 In the main dialog box. check all Equality of Parameters options. Click OK. All rights reserved. Test scale (Weibull or expo) or location (other dists) equal to: Enter a test value to compare the sample's scale (Weibull or exponential) or location (other distributions). scale. 3 To compare parameters from two or more distributions 1 2 In the main dialog box. Copyright © 2003–2005 Minitab Inc. 3 To test whether a sample comes from a historical distribution 1 2 In the main dialog box.

In Minimum X scale or Maximum X scale.. Hazard plot: Check to display a hazard plot.) • In Minimum X scale and Maximum X scale. Show graphs of different variables or by levels: Choose to display the graphs overlaid on the same graph or on separate graphs. If you like. To. or On separate graphs to have each plot displayed separately. enter values for the scale minimum and maximum. click Graphs. Click OK. All rights reserved. enter a label for the x-axis. Dialog box items Probability plot: Check to display a probability plot. Maximum of the tied points. • In X axis label. or Kaplan−Meier method. do any of the following: • Uncheck Display confidence intervals on above plots to turn off the 95. 4 To modify the default probability plot 1 2 In the main dialog box. modified Kaplan−Meier (Hazen). or Average (median) of tied points. enter the parameter of a historical distribution. Choose Overlaid on the same graph to have multiple samples plotted on the same graph.. To draw a parametric survival plot 1 2 3 In the main dialog box. (See To change confidence levels if you want to change from 95% confidence intervals. click Graphs. Parametric Distribution Analysis (Arbitrary Censoring) − Graphs Stat > Reliability/Survival > Distribution Analysis (Arbitrary Censoring) > Parametric Distribution Analysis > Graphs You can draw a probability plot. type values for the x-axis scale.0% confidence interval. Minimum X scale: Enter a minimum value to be contained within your scale. Survival plot: Check to display a survival plot.0% confidence interval Choose how plots are displayed For Parametric Distribution Analysis (Right Censoring) Under Handle tied failure times by plotting. . a survival plot.Reliability and Survivial Analysis • • 3 In Test scale (Weibull or expo) or location (other dists) equal to. Choose what to plot when you have tied failure times Turn off the 95. Maximum X scale: Enter a maximum value to be contained within your scale. choose All points (default). In Test threshold equal to. survival. In separate panels on the same graph to have multiple samples plotted in separate panels all on the same graph. Check Survival plot. Display confidence intervals on above plots: Check to display confidence intervals on the probability.. and cumulative failure plots. cumulative failure plot. enter the parameter of a historical distribution. Do any of the following: Do.. Click OK. X axis label: Enter text to identify your x-axis. type a label. and a hazard plot. See Tools > Options > Individual Graphs > Probability Plots to choose Median Rank (Benard). Cumulative failure plot: Check to display a cumulative failure plot. Mean Rank (Herd−Johnson). Uncheck Display confidence intervals on above plots. In X axis label. Specify the method used to obtain the plot points Specify a minimum and/or maximum value for the x-axis scale Enter a label for the x-axis 42 Copyright © 2003–2005 Minitab Inc.

The early period with high failure rate is often called the infant mortality stage. choose Least Squares (default) or Maximum Likelihood. Copyright © 2003–2005 Minitab Inc. low in the middle of the plot. Often. Each plot point represents the cumulative percentage of units failing at time t. The middle section of the curve. where failure rate increases again. which provide reasonable values for the "true" cumulative failure function. the curve often resembles the shape of a bathtub. Click OK. 43 . To change the method used to obtain the fitted line. In Confidence level. Cumulative failure plots Cumulative failure plots display the cumulative failure probabilities versus time. The end of the curve. click Estimate.0% confidence interval for the curve. Thus. This particular example does not have the bathtub shape.0% confidence interval to some other level. All rights reserved. is the normal life stage. enter a value. The cumulative failure curve is surrounded by two outer lines − the approximate 95. where the failure rate is low. the hazard rate is high at the beginning of the plot. click Estimate. In Estimation Method. Hazard plots − parametric distribution analysis The hazard plot displays the instantaneous failure rate for each time t. Click OK. then high again at the end of the plot.Distribution Analysis 3 4 5 Click OK. is the wearout stage. To change the confidence level for the 95.

Minitab provides two goodness of fit measures to help assess how the distribution fits your data. The percentiles may be transformed. . The Weibull probability plot below shows failure times associated with running engine windings at a temperature of 80° C: 44 Copyright © 2003–2005 Minitab Inc. Transformations of both the x and y data are needed to ensure that the plotted y values are a linear function of the plotted x values if the data are sampled from the particular distribution. see Parametric Distribution Analysis − Estimate. Fitted line. Minitab calculates the plot points using a nonparametric method. The plot consists of: • Plot points. the better the fit. To choose from various methods to obtain the fitted line. and are used as the x-variables. differ depending on the distribution used. In general. Confidence intervals. based on the chosen distribution. The associated probabilities are then transformed and used as the y-variables. the estimated cumulative probabilities (p) on the y-axis. chosen to linearize the fitted line. Minitab first calculates the percentiles for the various percents.Reliability and Survivial Analysis To draw a parametric hazard plot. The observed failure times are plotted on the x-axis vs. see Distribution ID Plot (Right Censoring) or Distribution ID Plot (Arbitrary Censoring). the closer the points fall to the fitted line. To make the fitted line.parametric distribution analysis. depending on the distribution. So you can use the probability plot to assess whether a particular distribution fits your data. The fitted line. differs depending on the parametric distribution chosen. The transformed scales. Tip To quickly compare the fit of up to eleven different distributions at once. set of approximate 95. • • For more information on probability plot calculations. see Tools > Options > Individual Graphs > Probability Plots. To choose from various methods to estimate the plot points. they would be the same (before being transformed) for any probability plot made. All rights reserved. Probability plots Use a probability plot to assess whether a particular distribution fits your data. which is a graphical representation of the percentiles. Because the plot points do not depend on any distribution. which represent the proportion of failures up to a certain time. however. see Methods and formulas . check Hazard plot in the Graphs subdialog box.0% confidence intervals for the fitted line.

which provide reasonable values for the "true" survival function.Distribution Analysis Survival plots Survival (or reliability) plots display the survival probabilities versus time. Copyright © 2003–2005 Minitab Inc. All rights reserved. 45 . The survival curve is surrounded by two outer lines − the approximate 95.0% confidence interval for the curve. Each plot point represents the proportion of units surviving at time t.

(See To change confidence levels if you want to change from 95% confidence intervals. In addition.0% confidence interval. choose All points (default). Show log-likelihood for each iteration of algorithm: Check to show the log-likelihood for each iteration of algorithm.. Check Survival plot. See Tools > Options > Individual Graphs > Probability Plots to choose Median Rank (Benard).0% confidence interval Choose how plots are displayed For Parametric Distribution Analysis (Right Censoring) Under Handle tied failure times by plotting. • In X axis label. parameter estimates. choose Least Squares (default) or Maximum Likelihood. enter values for the scale minimum and maximum. Display analyses for individual failure modes according to display of results: Check to display the results chosen above for each failure mode.) • In Minimum X scale and Maximum X scale.. goodness-of-fit and tests of parameters: Choose to display censoring information and the estimation method. If you like. Mean Rank (Herd−Johnson). To. Specify the method used to obtain the plot points Specify a minimum and/or maximum value for the x-axis scale Enter a label for the x-axis 3 4 5 Click OK. . All rights reserved. and percentiles. quartile information. click Estimate. If you select survival probabilities in the Estimate subdialog box. To change the confidence level for the 95. table of percentiles and survival probabilities: Choose to also display the MTTF. characteristics of distribution.. 46 Copyright © 2003–2005 Minitab Inc.Reliability and Survivial Analysis To draw a parametric survival plot 1 2 3 In the main dialog box. click Graphs. median. Do any of the following: Do. Choose Overlaid on the same graph to have multiple samples plotted on the same graph. Maximum of the tied points. enter a value. In Estimation Method. In Confidence level. Minitab also displays the assumed distribution. or Average (median) of tied points.. 4 To modify the default probability plot 1 2 In the main dialog box. log-likelihood. Dialog box items Control the Display of Results Display nothing: Choose to suppress all Session window output. Click OK. or Kaplan−Meier method. type values for the x-axis scale. Variable information. modified Kaplan−Meier (Hazen). and tests of parameters. Choose what to plot when you have tied failure times Turn off the 95. In X axis label.0% confidence interval to some other level. do any of the following: • Uncheck Display confidence intervals on above plots to turn off the 95. Parametric Distribution Analysis − Results Stat > Reliability/Survival > Distribution Analysis (Right Censoring) or Distribution Analysis (Arbitrary Censoring) > Parametric Distribution Analysis > Results You can control the display of Session window output. In Minimum X scale or Maximum X scale. Click OK. standard deviation. or On separate graphs to have each plot displayed separately. In separate panels on the same graph to have multiple samples plotted in separate panels all on the same graph. enter a label for the x-axis. Click OK. censoring information. To change the method used to obtain the fitted line. estimated parameters. type a label. Minitab also displays them. click Estimate. click Graphs. Uncheck Display confidence intervals on above plots.

and failure mode represents a unique distribution that requires estimates in one single column or in separate columns. enter columns of starting values based on your data structure. Use starting estimates: Enter starting estimates for the parameters. click Options. choose one of the following: • • 3 Estimate parameters of distribution. to enter your own estimates for the distribution parameters. Use historical estimates: Enter one column of values to be used for all samples. you can do any of the following: Starting estimates and historical estimates You must enter starting estimates or historical estimates for each distribution of failures each parameter you wish to estimate. To control estimation of the parameters 1 2 In the main dialog box. Enter a single column containing values for each parameter for all group levels. or several columns of values which match the order in which the corresponding variables appear in the Variables box in the main dialog box. Enter a single column containing values for each parameter for each variable. Dialog box items Estimate parameters of distribution: Choose to estimate the distribution parameters from the data (the default). Enter one column of values to be used for all samples. to estimate the distribution parameters from the data. each with a single failure mode. enter four columns containing parameter estimates. Two or more variables. In Maximum number of iterations. • • Enter one column containing values for each parameter of your distribution. If you choose Use historical estimates. Enter one column of values for each group level. Each variable. If Maximum Likelihood is your estimation method. If Least Squares is your estimation method. enter a positive integer to specify the maximum number of iterations for estimation. See Starting estimates for more information. Enter • • • A By variable with several group levels. Copyright © 2003–2005 Minitab Inc. By variable group level.Distribution Analysis Parametric Distribution Analysis − Options Stat > Reliability/Survival > Distribution Analysis (Right Censoring) or Distribution Analysis (Arbitrary Censoring) > Parametric Distribution Analysis > Options You can specify whether the distribution parameters should be calculated from the data or enter your own historical estimates. All rights reserved. 47 . If you choose Estimate parameters of distribution. Maximum number of iterations: Enter the maximum number of iterations. The default number of iterations is 20. See the following table for various scenarios: Variables with single failure modes: Case A single variable with a single failure mode. or several columns of values that match the order in which the corresponding variables appear in the Variables box in the main dialog box. Click OK. Use historical estimates. If there are four group levels. Enter one column of parameter estimates for each variable. choose Use historical estimates. See Historical estimates for more information. − Enter columns of historical values based on your data structure. − − 4 5 In Use starting estimates.

Enter a single column containing values for each parameter for each failure mode and variable. All rights reserved. If there are two variables with two failure modes. Confidence limits for percentiles: Check to store confidence limits for percentiles. enter four columns of parameter estimates. Variance/covariance matrix: Check to store the variance / covariance matrix. Available when you estimate survival probabilities at different times. Dialog box items Enter number of levels in by variable If all of the samples are stacked in one column. the first column should contain the parameter estimates of failure mode 1. variable 1. Normally. Survival probabilities: Check to store survival probabilities. . • • A By variable with several group levels and multiple failure modes. each with multiple failure modes. you can store information pertaining to the overall distribution only. Standard error of percentiles: Check to store standard error of percentiles. If there are three failure modes. Characteristics of Fitted Distribution Percentiles: Check to store percentiles. Percents for percentiles: Check to store percents for percentiles. Information on Parameters Choose to store the following when you have a single failure mode. Enter a single column containing values for each parameter for each failure mode of each group level. If there are two group levels with three failure modes each. Available when you estimate survival probabilities at different times. If variable 2 does not have failure mode 1.Reliability and Survivial Analysis Variables with multiple failure modes: Case A single variable with a multiple failure modes. Enter one column of parameter estimates for each failure mode. enter the number of levels the column of grouping indicators contains. Available when you estimate survival probabilities or cumulative failure probabilities at different times. Cumulative failure probabilities: Check to store cumulative failure probabilities. Log-likelihood for last iteration: Check to store the log-likelihood ratio for the last iteration. the second column contains the parameter estimates of failure mode 2. enter six columns of parameter estimates. Two or more variables. Parameters estimates: Check to store parameter estimates. variable 1. variable 2. Enter one column of parameter estimates for each failure mode of each variable. • • Parameter estimates are assigned by failure mode then variable. Confidence limits for parameters: Check to store confidence limits for parameters. Enter one column of parameter estimates for each failure mode of each group level. Caution Do not enter starting estimates or historical estimates for failure modes that you have eliminated from the analysis. Times for probabilities: Check to store times for survival probabilities and cumulative failure probabilities. Confidence limits for cumulative failure probabilities: Check to store confidence limits for cumulative failure probabilities. With multiple failure modes. Available when you estimate cumulative failure probabilities at different times. 48 Copyright © 2003–2005 Minitab Inc. the second column − parameter estimates of failure mode 1. Standard error of estimates: Check to store standard error of parameter estimates. enter three columns of parameter estimates. Enter • • Enter a single column containing values for each parameter for each failure mode. Parametric Distribution Analysis − Storage Stat > Reliability/Survival > Distribution Analysis (Right Censoring) or Distribution Analysis (Arbitrary Censoring) > Parametric Distribution Analysis > Storage You can store characteristics of the fitted distribution and information on the distribution parameters in your worksheet. Available when you estimate cumulative failure probabilities at different times. etc. Confidence limits for survival probabilities: Check to store confidence limits for survival probabilities.

3 20482.993 813.593 594. In Frequency columns.000 miles) to see if the tire has failed.8 Copyright © 2003–2005 Minitab Inc. then click OK. In Estimate probabilities for these times (values). Check Survival plot.1 Parameter Location Scale Estimate 78016.4 81723. Click Estimate.4 Log-Likelihood = -1468. Session window output Distribution Analysis.6 84055.5 71301. 1 2 3 4 5 6 7 Open the worksheet TIREWEAR. enter End." or wear down to 2/32 of an inch of tread.998 Characteristics of Distribution Standard Error 685. You are interested in finding out how many miles it takes for various proportions of the tires to "fail.932 95.9 21920.0% Normal CI Lower Upper 68059.838 635.2 14920. Choose Stat > Reliability/Survival > Distribution Analysis (Arbitrary Censoring) > Parametric Distribution Analysis In Start variables. enter 45000.325 Correlation Coefficient = 0. Click Graphs.MTW.Distribution Analysis Example of a parametric distribution analysis with arbitrarily censored data Suppose you work for a company that manufactures tires. You are especially interested in knowing how many of the tires last past 45.0 79168. In End variables.6 70748. All rights reserved. enter Start.6 Mean(MTTF) Standard Deviation Median First Quartile(Q1) Third Quartile(Q3) Interquartile Range(IQR) Estimate 69403.000 miles.7 23462.7 15970.8 57632. choose Smallest extreme value.9 82889. then enter the data into the Minitab worksheet. Click OK in each dialog box.4 13939.0% Normal CI Lower Upper 76864. From Assumed distribution.4 61221.9 19136.757 915. enter Freq.3 17878.2 72547. You inspect each tire at regular intervals (every 10.6 73793.7 59426. 49 .866 517.593 95.900 663.6 25113. Start = Start and End = End Variable Start: Start Frequency: Freq Censoring Information Right censored value Interval censored value Left censored value End: End Count 71 694 8 Estimation Method: Least Squares (failure time(X) on rank(Y)) Distribution: Smallest Extreme Value Parameter Estimates Standard Error 587.686 Goodness-of-Fit Anderson-Darling (adjusted) = 2.

4 81934.4 24031.5 100802 Table of Survival Probabilities 95.6 66597.16 19797.855 799.5 41780.70 2159.2 92868.0 35872.7 33844.1 15564.7 26741.0 38073.5 41882.4 93251.0% Normal CI Lower Upper 4461.18 1812.5 37033.67 1609.2 95107.4 94927.8 44439.1 64267.2 38877.3 43824.88 1020.90 1956.5 95905.3 79637.86 1533.5 62634.9 80785.919 693.1 30366.722 936.Reliability and Survivial Analysis Table of Percentiles Standard Error 2509.4 53636.9 61001.594 871.9 91840.4 97026.4 92609.757 594.6 40025.5 45548.8 77876.934 774.9 92511.2 91166.004 735.8 33699.4 89101.11 1700.877714 0.3 91127.0% Normal CI Lower Upper 0.8 85116.6 30292.4 83913.162 585.7 89744.872 720.8 94386. .896380 50 Copyright © 2003–2005 Minitab Inc.1 93449.9 98368.7 95458.4 90429.8 55636.974 613.170 705.0 39653.27 1467.1 22089.84 1356.021 712.741 830.8 100077 102637 Percent 1 2 3 4 5 6 7 8 9 10 20 30 40 50 60 70 80 90 91 92 93 94 95 96 97 98 99 Percentile 9380.6 90460.6 94051.5 33343.3 57636.6 75547.3 67993.6 96735.048 95.0 47099.9 91818.1 40949.8 72547.7 25923.6 73793.3 86319.6 29757.7 76711.0 71301.8 91972.1 98363.876 753.7 36498.0 93891.9 98967.9 96659.912340 Time 45000 Probability 0.08 1408.24 14299.0 69390. All rights reserved.23 833.680 635.1 42786.

7 miles.Distribution Analysis Graph window output Interpreting the results As shown in the Characteristics of Distribution table. All rights reserved. 51 . respectively.9 and 72.403. the mean and median miles until the tires fail are 69. Copyright © 2003–2005 Minitab Inc.547.

26856 57.4415 74. you can see that 89.90747 3.1075 52 Copyright © 2003–2005 Minitab Inc.2497 15.0% Normal CI Lower Upper 3.0% Normal CI Lower Upper 51.7 miles and 50% fail by 72.5866 68.Reliability and Survivial Analysis To see the times at which various percentages or proportions of the tires fail.78798 1. For example.547.7777 44.258 Goodness-of-Fit Anderson-Darling (adjusted) = 67.4674 68.84080 95.7956 Log-Likelihood = -202. Output − Parametric Distribution Analysis The default output for Parametric Distribution Analysis (Right Censoring) and Parametric Distribution Analysis (Arbitrary Censoring) is: • • Censoring information Parameter estimates and their .Log-likelihood and goodness-of-fit statistics Characteristics of distribution and their .Standard errors .986 Characteristics of Distribution Standard Error 2.95% confidence intervals Probability plot for each failure mode • • • For example.7720 39.6179 17.95% confidence intervals Table of percentiles and their .7498 Mean(MTTF) Standard Deviation Median First Quartile(Q1) Third Quartile(Q3) Interquartile Range(IQR) Estimate 56. .7 miles.699.Standard errors .00877 2.27922 95.606 Correlation Coefficient = 0.6177 Parameter Shape Scale Estimate 3.6757 19.4955 62. Distribution Analysis: Temp80 Variable: Temp80 Censoring Information Uncensored value Right censored value Count 37 13 Censoring value: Cens80 = 0 Estimation Method: Least Squares (failure time(X) on rank(Y)) Distribution: Weibull Parameter Estimates Standard Error 0.7263 26.3817 21.000 miles.10091 4.6941 24. In the Table of Survival Probabilities.0996 51. look at the Table of Percentiles.3558 62.3032 56.0629 50.64% of the tires last past 45. 5% of the tires fail by 33.Standard errors .8914 63.73936 0. All rights reserved.872053 2.296611 2.63819 62.95% confidence intervals .

81118 2.3005 52.79777 2.6375 97.81001 2. Copyright © 2003–2005 Minitab Inc.95707 2.7720 55.3384 77.5499 From the Results subdialog box.8233 66.7796 53.1830 36. This graph is often used before the full analysis to view summary information.2256 30.1636 83.2431 16.0423 27.5024 39.9618 24.8274 26. and draw survival and hazard plots.1542 28.93652 3.85714 2.5509 46. and create a Distribution Overview Plot.2159 73.7094 84.6817 76. 53 .82909 2.90747 2.5920 88.4656 89.0% Normal CI Lower Upper 13.3374 19. on one graph. use Parametric Distribution Analysis (Arbitrary Censoring).7339 21.6162 74.Right Censoring also tests the equality of their survival curves. hazard rates.3558 62.8624 78.6000 85.9283 41.1794 79.0728 35.7063 86.97051 3.89222 2. All rights reserved. You can enter up to 50 columns (50 different samples).8866 27. and other estimates depending on the nonparametric technique chosen.78019 2.9520 81.16895 95. Nonparametric Distribution Analysis .0809 94. When you have multiple samples. You can estimate survival probabilities. Dialog box items Start variables: Enter the columns of start times. hazard estimates.1142 92.0717 47. Command Distribution Overview Plot Right Censored Arbitrary Censored Nonparametric Distribution Analysis Right Censored Arbitrary Censored Description Draws a Kaplan-Meier survival plot and hazard plot on one graph. Nonparametric Distribution Analysis (Arbitrary Censoring) Stat > Reliability/Survival > Distribution Analysis (Arbitrary Censoring) > Nonparametric Distribution Analysis Use this command when you have arbitrarily censored data or actual failure times and no distribution fits your data.80866 2.6707 85. cumulative failure probabilities.5708 78.4461 84.8008 71.7259 58.Distribution Analysis Table of Percentiles Standard Error 2.2342 30. End variables: Enter the columns of end times. You can enter up to 50 columns (50 different samples).5305 23.2010 22. and draw survival.3046 66.0525 26.3603 95. or an Actuarial survival plot and hazard plot.78771 2.9748 79.4865 90.0685 27. and other functions.6859 33.4531 38.8345 87.0084 82.95551 2.7777 61.8152 33.7228 32.74337 2.6536 91. You can request Turnbull or Actuarial estimates.1039 32.85336 2.64045 2.94029 2.01533 3. and hazard plots.85684 2.1676 60.9534 67.4857 24.8299 41. Gives you nonparametric estimates of the survival probabilities. cumulative failure. which perform the full analysis. cumulative failure probabilities.5795 47.77225 2.0778 21. Frequency columns (optional): Enter the columns of frequency data.2089 56.3174 36. Draws a Turnbull survival plot.99523 2.44778 2.967 Percent 1 2 3 4 5 6 7 8 9 10 20 30 40 50 60 70 80 90 91 92 93 94 95 96 97 98 99 Note Percentile 17.3351 51.0826 71.6796 81.4813 25.5364 101.4812 40.0814 89.4397 83. you can request additional output and display Session window output for each failure mode.6068 75. If a distribution fits your data.91916 2.7571 29.5348 31.8988 86.88716 2. Nonparametric Distribution Analysis Nonparametric distribution analysis commands The nonparametric distribution analysis commands include Nonparametric Distribution Analysis-Right Censoring and Nonparametric Distribution Analysis-Arbitrary Censoring.01453 2.

Exact failures at 30000 hours. Start * 10000 20000 30000 30000 40000 50000 50000 60000 70000 80000 90000 End 10000 20000 30000 30000 40000 50000 50000 60000 70000 80000 90000 * Right censored at 90000 hours. For observations with corresponding columns of frequency. . Left censored at 10000 hours. All rights reserved. For systems that have more than one cause of failure. Failure time Time that the failure occurred after Missing value symbol '∗' Enter in the End Column.. 54 Copyright © 2003–2005 Minitab Inc.. check By variable. left. Alternatively. If all of the samples are stacked in one column. You can enter up to 50 samples per analysis. see Multiple Failure Modes (Arbitrarily Censored Data). your data is arbitrarily-censored. and enter a column of grouping indicators in the box. Enter your data in table form. using a Start column and End column: For this observation.. see Using frequency columns. including right-. To do a nonparametric distribution analysis (arbitrarily censored data) 1 2 3 4 5 Choose Stat > Reliability/Survival > Distribution Analysis (Arbitrary Censoring) > Nonparametric Distribution Analysis.. All the samples display on a single plot. enter up to 50 columns of start times. For an illustration. check By variable. unless you assume a common shape (Weibull) or scale (other distributions). Interval censored between 50000 and 60000 hours. enter them in Frequency columns. see Stacked vs. then set up a column of grouping indicators. When you have more than one sample. Minitab analyzes systems with one cause of failure or multiple causes of failure. which helps you compare the various functions between samples. When you have frequency columns. see Distribution Analysis Data. Exact failure time Right censored Left censored Interval censored Enter in the Start Column.Reliability and Survivial Analysis By variable: If all of the samples are stacked in one column. Failure time Missing value symbol '∗' Time before which the failure occurred Time at start of interval during which the Time at end of interval during which the failure occurred failure occurred This data set illustrates tabled data. which can be numbers or text. with different colors and symbols. you can stack all the samples in one column. Minitab estimates the functions independently for each sample. In Start variables. Distribution Analysis (Arbitrarily Censored Data) When your data consist of exact failures and a varied censoring scheme. and enter a column of grouping indicators in the box. enter up to 50 columns of end times. For general information on life data and censoring. In End variables. Unstacked data. you can use separate columns for each sample...and interval-censored data.

Dialog box items Use failure mode columns: Enter the columns containing the failure modes. and Screen) to see which component to improve. To display Session window output for all of the failure modes. Nonparametric Distribution Analysis − Failure Mode Stat > Reliability/Survival > Distribution Analysis (Right Censoring) or Distribution Analysis (Arbitrary Censoring) > Nonparametric Distribution Analysis > FMode Use to estimate the overall reliability of your system when multiple causes of failure exist in order to investigate the reliability of the individual failure modes. In End variables. You have interval failure times and are unsure of the distribution of failures. Click OK. In Frequency columns (optional). enter End. To be competitive in your market. Right endpoint of intervals: Choose if the experiment continues until the right endpoint when a failure in the interval occurs. Failure Mode Options Use all failure modes: Choose to include all failure modes in the analysis. Hinge. Window Censoring Information Right censored value Interval censored value Left censored value Count 18 55 7 Copyright © 2003–2005 Minitab Inc. enter Start. In Use failure mode columns. then check Display analyses for individual failure modes according to display of results. You decide to look at each of the three failure modes (Window. In Start variables.000 cycles should be at least 90%. Session window output Distribution Analysis. Check Survival plot. Choose Stat > Reliability/Survival > Distribution Analysis (Arbitrary Censoring) > Nonparametric Distribution Analysis. use any of the dialog box options. Create right censored observations using (Only available with arbitrary censoring.MTW.) Use to determine how Minitab will create right-censored observations for other failure modes when data are interval censored. Click Graphs. Click OK in each dialog box. Start = Start and End = End Variable Start: Start Frequency: Freq End: End Failure Mode: Failure = Hinge. enter Freq. click Results. 55 . Example of nonparametric distribution analysis with multiple failure modes You work for a medical imaging company and are designing a new x-ray cassette. Use failure modes: Enter the failure modes to include in the analysis. then click OK. enter Failure. Midpoint of intervals: Choose if a failure for any one failure mode causes the experiment to end. Eliminate failure modes: Enter the failure modes to exclude from the analysis.Distribution Analysis 6 If you like. Click FMode. All rights reserved. This option is available for both right-censored and arbitrarily censored data. Screen. the overall cassette reliability at 20. 1 2 3 4 5 6 7 8 Tip Open the worksheet CASSETTE.

0335410 0.2250 Standard Error 0.5125 0.0124216 * Time 2500 5000 7500 10000 12500 15000 17500 20000 22500 25000 27500 30000 32500 Survival Probability 0.2375 0.428243 0. .0270633 0.353243 0.492648 0.6250 0.3625 0. All rights reserved.6000 0.133495 0.683495 0.571757 0.0335410 0.0315919 0.707352 0.0500 0.0475781 0.5375 0.402969 0.Reliability and Survivial Analysis Turnbull Estimates Interval Lower Upper * 2500 2500 5000 5000 7500 7500 10000 10000 12500 12500 15000 15000 17500 17500 20000 20000 22500 22500 25000 25000 27500 27500 30000 30000 32500 32500 * Probability of Failure 0.0294480 0.0875 0.0537464 0.6750 0.0518223 0.0174553 0.777635 0.4625 0.646757 0.0558318 0.850581 0.974419 0.0174553 0.1000 0.0557443 0.0243670 0.0541266 0.365572 0.1375 0.0750 0.210930 0.316505 Graph window output 56 Copyright © 2003–2005 Minitab Inc.0466871 95.0250 0.0125 0.3125 0.0375 0.1000 0.0124216 0.144249 0.0547723 0.622031 0.866505 0.2250 Standard Error 0.330751 0.0315919 0.518914 0.0243670 0.0% Normal CI Lower Upper 0.4750 0.0557443 0.7750 0.731086 0.0385022 0.572365 0.0500 0.0558842 0.0250 0.0523659 0.0212408 0.0466871 0.0625 0.584428 0.0125 0.257159 0.414070 0.467841 0.9125 0.

and enter a series of numbers. and hazard plots. With Nonparametric Distribution Analysis (Arbitrary Censoring). For example. etc. Confidence level: Enter a confidence level for all of the confidence intervals. 4 Click OK.4750 and you are 95% confident that the true reliability is between 0. Nonparametric Distribution Analysis − Graphs Stat > Reliability/Survival > Distribution Analysis (Right Censoring) or Distribution Analysis (Arbitrary Censoring) > Nonparametric Distribution Analysis > Graphs You can draw survival plots.9744. density (actuarial method) estimates and log-rank and Wilcoxon statistics.0%. and so on up to 80-100. or a column of numbers. For example. Do one of the following and then click : With Nonparametric Distribution Analysis (Right Censoring). in the box. you determine that the 20. choose In addition. hazard and density estimates (actuarial method). click Estimate. In Confidence level.000-cycle reliability for the components are: Hinge=0. The default is 95. Significant improvements to the hinge and window are required to meet the overall cassette reliability target.6516. 10-20. you need to understand each component's reliability for improvement. 20-40. gives you these time intervals: 0-4. enter a value. click Estimate. use unequally spaced time intervals − choose Enter endpoints of intervals. Because the actual reliability is much worse than your target. Dialog box items Estimation Method Turnbull: Choose to estimate the parameters using the Turnbull method. 8-10.584428. All rights reserved. Window=0. check Actuarial. cumulative failure probabilities. By examining the survival plots. Click OK. Estimate survival probabilities: Choose to estimate the survival probabilities. 68. hazard. and Screen=0. Under Estimation Method. do one of the following: • • use equally spaced time intervals − choose 0 to_by_ and enter numbers in the boxes. 0 to 100 by 20 gives you these time intervals: 0-20. With Nonparametric Distribution Analysis (Right Censoring). and 20-30. Copyright © 2003–2005 Minitab Inc. More To change confidence levels 1 2 3 In the main dialog box. Nonparametric Distribution Analysis (Arbitrary Censoring) − Estimate Stat > Reliability/Survival > Distribution Analysis (Arbitrary Censoring) > Nonparametric Distribution Analysis > Estimate You can choose the method of estimation for the nonparametric estimates of survival probabilities.73574. Dialog box items Survival plot: Check to display a survival plot. 4-6.Distribution Analysis Interpreting the results The overall 20.365572 and 0. entering 0 4 6 8 10 20 30. To display hazard and density estimates in the Actuarial table. 57 . Actuarial method: Choose to estimate the parameters using the actuarial method. choose In addition. To request actuarial estimates 1 2 3 In the main dialog box. Estimate cumulative failure probabilities: Choose to estimate the cumulative failure probabilities. click Results. Confidence intervals: Choose to use two-sided confidence intervals (the default) or just an upper or lower confidence interval.000-cycle cassette reliability is 0. cumulative failure plots. hazard rates. from the main dialog box.

Reliability and Survivial Analysis Cumulative failure plot: Check to display a cumulative failure plot. Click OK. To draw a hazard plot 1 2 3 4 In the Nonparametric Distribution Analysis (Arbitrary Censoring) dialog box. Often. Since the Actuarial method is not the default estimation method. All rights reserved. click Estimate. Check Hazard plot. be sure to choose Actuarial method in the Estimate subdialog box when you want to draw a hazard plot. Choose Actuarial. Maximum X scale: Enter a value for the maximum X-axis scale. then click OK. If you like. . Hazard plots − nonparametric distribution analysis The hazard plot displays the instantaneous failure rate for each time t. low in the middle of the plot. the hazard rate is high at the beginning of the plot. The end of the curve. You can optionally plot Actuarial estimates. Nonparametric hazard estimates are calculated various ways: • • Nonparametric Distribution Analysis (Right Censoring) automatically plots the empirical hazard function. 58 Copyright © 2003–2005 Minitab Inc. is the wearout stage. enter the desired text in this box. Nonparametric Distribution Analysis (Arbitrary Censoring) only plots Actuarial estimates. Click Graphs. X axis label: To replace the default X-axis label with your own label. where the failure rate is low. Thus. then high again at the end of the plot. Hazard Plots: Check to display a hazard plot. Show graphs of different variables or by levels: Choose to display the graphs overlaid on the same graph or on separate graphs. where failure rate increases again. The middle section of the curve. the curve often resembles the shape of a bathtub. Display confidence intervals on plot: Check to display confidence intervals on the survival and cumulative failure plots. is the normal life stage. Minimum X scale: Enter a value for the minimum X-axis scale. The early period with high failure rate is often called the infant mortality stage. use any dialog box options.

You can interpret the nonparametric survival curve in a similar manner as you would the parametric survival curve. In addition. Nonparametric Distribution Analysis (Arbitrary Censoring) − Results Stat > Reliability/Survival > Distribution Analysis (Arbitrary Censoring) > Nonparametric Distribution Analysis > Results You can control the display of Session window output. In addition. depending on the command you use: • • With Nonparametric Distribution Analysis (Right Censoring). the survival plot uses Turnbull survival estimates by default. but you can choose to plot Actuarial estimates. Display analyses for individual failure modes according to display of results: Check to display the results chosen above for each failure mode. The survival curve is surrounded by two outer lines − the 95% confidence interval for the curve. 59 . Each plot point represents the proportion of units surviving at time t. hazard and density estimates (actuarial method): Choose to also display hazard estimates and density estimates. The major difference is that the nonparametric survival curve is a step function while the parametric survival curve is a smoothed function. Variable information. Copyright © 2003–2005 Minitab Inc. the survival plot uses Kaplan−Meier survival estimates by default. censoring information and characteristics of variable (actuarial method): Choose to display censoring information and an additional time table. See Comparing survival curves when you perform nonparametric distribution analysis on more than one right−censored sample. You can choose from various estimation methods.Distribution Analysis Nonparametric survival plots Survival (or reliability) plots display the survival probabilities versus time. but you can choose to plot Actuarial estimates. All rights reserved. Turnbull survival probabilities or actuarial table: Choose to also display Turnbull survival probabilities or actuarial tables. Dialog box items Control the display of results Display nothing: Choose to suppress all Session window output. which provide reasonable values for the "true" survival function. With Nonparametric Distribution Analysis (Arbitrary Censoring).

Example of nonparametric distribution analysis with arbitrarily censored data Suppose you work for a company that manufactures tires. Available when you choose the Actuarial estimation method. . then click OK. Available when you choose the Actuarial estimation method. within given mileage intervals. You are interested in finding out how likely it is that a tire will "fail. Cumulative failure probabilities: Check to store cumulative failure probabilities.012937 0. Survival probabilities: Check to store survival probabilities.187581 0. All rights reserved. Dialog box items Enter number of levels in by variable: If all of the samples are stacked in one column. In Frequency columns.0076784 0.018111 0. Hazard rates: Check to store the hazard estimates.0047964 0.0164640 0.010349 0.MTW.0140409 0. Standard error for cumulative failure probabilities: Check to store standard errors for cumulative failure probabilities.091850 Standard Error 0.187581 0. 1 2 3 4 Open the worksheet TIREWEAR.047865 0. Confidence limits for cumulative failure probabilities: Check to store confidence limits for cumulative failure probabilities. Standard error for survival probabilities: Check to store standard errors for survival probabilities.0113672 0. Session window output Distribution Analysis. In Start variables.112549 0. Confidence limits for survival probabilities: Check to store confidence limits for survival probabilities.298836 0. enter Start.0063628 0. then enter the data into the Minitab worksheet." or wear down to 2/32 of an inch of tread.0036400 0.000 miles) to see if the tire fails. enter Freq. You are especially interested in knowing how many of the tires last past 45. Times for hazard rates: Check to store the times for the hazard estimates.Reliability and Survivial Analysis Nonparametric Distribution Analysis (Arbitrary Censoring) − Storage Stat > Reliability/Survival > Distribution Analysis (Arbitrary Censoring) > Nonparametric Distribution Analysis > Storage You can store various probability and rate estimates. Start = Start and End = End Variable Start: Start Frequency: Freq Censoring Information Right censored value Interval censored value Left censored value End: End Count 71 694 8 Turnbull Estimates Interval Lower Upper * 10000 10000 20000 20000 30000 30000 40000 40000 50000 50000 60000 60000 70000 70000 80000 80000 90000 90000 * Probability of Failure 0. Nonparametric estimates Check any of the items below to store them in the worksheet.0040644 0. enter the number of levels the column of grouping indicators contains.032342 0. In End variables.0140409 * 60 Copyright © 2003–2005 Minitab Inc. You inspect each good tire at regular intervals (every 10. Choose Stat > Reliability/Survival > Distribution Analysis (Arbitrary Censoring) > Nonparametric Distribution Analysis. Times for probabilities: Check to store times for survival probabilities and cumulative failure probabilities. enter End.000 miles.

989651 0.926261 0.311063 0.0036400 0.7 Median = 55 IQR = * Q1 = 48 Q3 = * Copyright © 2003–2005 Minitab Inc.0254 Mean(MTTF) 55.0152311 0.987345 0.0093999 0.247798 0. interquartile range. including the mean.3746 60. its standard error and 95% confidence intervals.944560 0.0% Normal CI Lower Upper 51.112210 Interpreting the results The Turnbull Estimates table displays the probabilities of failure. 61 .000 and 70.63% of the tires are estimated to survive until 40.996785 0.95% confidence intervals Default estimation method Kaplan−Meier Turnbull Optional estimation method Actuarial Actuarial For example. Output − Nonparametric Distribution Analysis The default output for Nonparametric Distribution Analysis depends on the estimation method you choose: Data Right−censored Arbitrarily−censored Kaplan−Meier estimates • • • Censoring information Characteristics of the variable. median. Q1.Standard error .76% of the tires are estimated to fail between 60.795700 0.000 miles.0117552 0.958603 0.0177621 0.765847 0.071490 0.0161393 0.0071650 0.944685 0.543453 0.20686 95.613079 0. 18.855356 0.091850 Standard Error 0.0% Normal CI Lower Upper 0.735995 0. and Q3 Kaplan-Meier estimates of survival probabilities and their .0054243 0. All rights reserved.Distribution Analysis Time 10000 20000 30000 40000 50000 60000 70000 80000 90000 Survival Probability 0.279431 0. Distribution Analysis: Temp80 Variable: Temp80 Censoring Information Uncensored value Right censored value Count 37 13 Censoring value: Cens80 = 0 Nonparametric Estimates Characteristics of Variable Standard Error 2.000 miles.878396 0.578266 0.0103879 95. You can see in the column of survival probabilities that 92.976714 0.966083 0.907838 0.972646 0.901436 0.982516 0. For example.

0140409 0.0459565 0.258462 Standard Error 0.789927 0.689128 0.0634980 0.0140409 * 62 Copyright © 2003–2005 Minitab Inc.0678823 0.38029 Turnbull estimates • • • Censoring information Turnbull estimates of the probability of failure and their standard errors Turnbull estimates of the survival probabilities and their standard errors and 95% confidence intervals For example.641621 0.800000 0.665179 0.51454 0.0490714 0.905684 1.136632 0.360000 0.0701997 0.226953 0.63859 0.381521 0.440000 0.941195 1.0197990 0.032342 0.550702 0.361410 0.680000 0.0697997 0.0701997 0.047865 0.0669925 0.500000 0.700000 0.0659697 0.92649 0.340000 0.0040644 0.61848 0.280000 0.780000 0.155546 0.572980 0.89482 0.012937 0.0706541 0.87838 0.47130 0.0603987 0.97007 0.0063628 0.95618 0. Start = Start and End = End Variable Start: Start Frequency: Freq Censoring Information Right censored value Interval censored value Left censored value End: End Count 71 694 8 Turnbull Estimates Interval Lower Upper * 10000 10000 20000 20000 30000 30000 40000 40000 50000 50000 60000 60000 70000 70000 80000 80000 90000 90000 * Probability of Failure 0.302411 0.401854 0.82702 0.960000 0.91087 0.091850 Standard Error 0.0047964 0.57759 0.0565685 0.860000 0.0518459 0.94162 0.443195 0.00000 0.560000 0.840000 0.0669925 0.0706541 0.71680 0.187581 0. .65848 0.190702 0.400000 0.79130 0.0686440 0.380000 0.0585833 0.80930 0. All rights reserved.208697 0.980000 0.0113672 0.264210 0.0707107 0.0036400 0.320000 0.0704840 0.713511 0.920000 0.112549 0.0648074 0.010349 0.0277128 0.0% Normal CI Lower Upper 0.540000 0.880000 0.99520 0.420000 0.844803 0.0424264 0.283195 0.44930 0.98315 0.0697997 0.40445 0.49305 0.760000 0.816846 0.660000 0.0164640 0.00000 0.55680 0.53579 0.298836 0.0621592 95.763822 0.738384 0. Distribution Analysis.018111 0.341521 0.900000 0.0076784 0.520000 0.422411 0.528697 0.0659697 0.67815 0.0692820 0.580000 0.480000 0.820000 0.0543323 0.187581 0.245460 0.0383667 0.69759 0.Reliability and Survivial Analysis Kaplan-Meier Estimates Number at Risk 50 49 48 46 45 44 43 42 41 40 39 38 35 34 33 29 28 27 26 25 24 22 21 20 19 18 17 16 13 Time 23 24 27 31 34 35 37 40 41 45 46 48 49 50 51 52 53 54 55 56 58 59 60 61 62 64 66 67 74 Number Failed 1 1 2 1 1 1 1 1 1 1 1 3 1 1 4 1 1 1 1 1 2 1 1 1 1 1 1 2 1 Survival Probability 0.

0152311 0.000000 0.0054243 0.878396 0.0161393 0.7900 Median 56. we requested equally spaced time intervals from 0-110.976714 0.071490 0.578266 0.0% Normal CI Lower Upper 0.9514 26. In this example.0000 Standard Error 3.160000 0.36718 95.944685 0.00 0.0117552 0.765847 0.613079 0.0% Normal CI Lower Upper 49. you can request Actuarial estimates in the Estimate subdialog box.091850 Standard Error 0. in increments of 20: Distribution Analysis: Temp80 Variable: Temp80 Censoring Information Uncensored value Right censored value Count 37 13 Censoring value: Cens80 = 0 Nonparametric Estimates Characteristics of Variable Standard Error 3.051846 0.0103879 95.08607 95.996785 0.421053 0. you can request specific time intervals.Distribution Analysis Time 10000 20000 30000 40000 50000 60000 70000 80000 90000 Survival Probability 0.982516 0.0071650 0.0177621 0.966083 0.112210 Actuarial survival estimates Instead of the default Kaplan-Meier or Turnbull survival estimates. All rights reserved.000000 0.000000 0.926261 0.500000 0.972646 0.077152 0.907838 0. 63 .311063 0.36718 3.5909 42.247798 0.855356 0.113269 0.000000 Interval Lower Upper 0 20 20 40 40 60 60 80 80 100 100 120 Number Entering 50 50 42 21 9 3 Number Failed 0 8 21 8 0 0 Number Censored 0 0 0 4 6 3 Standard Error 0.000000 0.5909 62.1905 20.989651 0. • • • Median residual lifetimes Conditional probabilities of failure Survival probabilities With Nonparametric Distribution Analysis-Right Censoring.901436 0.795700 0.987345 0.543453 0.7900 13.84 Time T 20 40 Additional Time 36.279431 0.0% Normal CI Lower Upper 29.0486 Actuarial Table Conditional Probability of Failure 0.0093999 0.0036400 0.735995 0.1905 Additional Time from Time T until 50% of Running Units Fail Proportion of Running Units 1.944560 0.958603 0.000000 Copyright © 2003–2005 Minitab Inc.

12082 0. 64 Copyright © 2003–2005 Minitab Inc.0518459 0. with higher values indicating a better fitting distribution.24316 0.24316 0.0% Normal CI Lower Upper 1.73838 0.12082 0.84000 0.94162 0. you must also enter a column defining the failure mode. .00000 1. but required for multiply censored data. C designates a unit that was removed from the test before failure. The Censor column is optional for singly censored data.55680 0. . and so on The Censor column contains the corresponding censoring indicators: F designates an actual failure. .0697997 0. A smaller Anderson−Darling statistic indicates that the distribution fits the data better. see Using frequency columns. See Multiple Failure Modes (Right Censored Data). . Worksheet Structure (Uncensored/Right-Censored Data) To perform reliability studies. The correlation will range between 0 and 1. The Pearson correlation measures the strength of the linear relationship between the X and Y variables on a probability plot. and so on Multiply censored data Months 50 53 60 65 70 70 50 53 .0000000 0.Reliability and Survivial Analysis Time 20 40 60 80 100 120 Note Survival Probability 1. • • Anderson−Darling statistic for the maximum likelihood and least squares estimation methods.36550 From the Results subdialog box. . Singly censored data You can define the censoring in one of the following ways: • Censoring columns − Enter two columns for each sample. .0624194 0. you can request additional output and display Session window output for each failure mode.28320 0. .00000 0. Right Censored Data Goodness-of-fit statistics Minitab displays up to two goodness−of−fit statistics to help you compare the fit of distributions. because the statistic changes when a different plot point method is used.00000 0.42000 0.0624194 95. . and so on Censor F F C C F F F F .36550 0. For systems with more than one cause of failure.12082 0. Minitab uses an adjusted Anderson−Darling statistic.0624194 0.36550 0. For observations with corresponding columns of frequency. Worksheet structure : Singly censored data Months 50 50 53 53 60 65 70 70 . The statistic is a weighted squared distance from the plot points to the fitted line with larger weights in the tails of the distribution. Pearson correlation coefficient for the least squares estimation method. enter a column showing the failure times of each sample. The Anderson−Darling statistic is a measure of how far the plot points fall from the fitted line in a probability plot. and so on Note Censor F F F F F F C C . The columns for each sample must be the same length. .24316 Standard Error 0. although pairs of columns from different samples can have different lengths. one column of failure times and a corresponding column of censoring indicators. All rights reserved.

Failure censoring means that you run the study until you observe a specified number of failures. Freq 1 4 1 35 42 1 39 1 . Censor F F C F F C F F . units are failure censored at 7. etc. warranty data usually includes large numbers of observations with common censoring times. . This is known as Type I censoring on the right. Four failures at 150 days. etc. Censor F F F F F C F F . Using frequency columns (uncensored/right censored data) You can structure each column so that it contains individual observations (one row = one observation) or unique observations with a corresponding column of frequencies (counts). grouping indicators can be numbers or text. Minitab assumes the lower of the two values indicates censoring. etc. All rights reserved. In the example above. • Failure Censoring − Enter a column of failure times and the number of failures at which to begin censoring. Time censoring means that you run the study for a specified period of time. This is known as Type II censoring on the right. . For example. Unstacked data In unstacked data. you can stack all the data in one column and add a column of grouping indicators that define each sample. Frequency columns Days 140 150 151 151 153 161 170 199 . Here are the same data structured both ways: Individual observations Days 140 150 150 150 150 151 151 151 . 65 . All units still running at the end time are time censored. Copyright © 2003–2005 Minitab Inc. You specify the number of failures at which to begin censoring. . If you do not enter a censoring value in the Censor subdialog box. Multiply censored data You must define the censoring using censoring columns. . etc. each sample is in a separate column. Like censoring indicators. Alternatively. Stacked vs. . units are time censored at 70 months. Frequency columns are useful for data where you have large numbers of observations with common failure and censoring times. • Time Censoring − Enter a column of failure times and the time at which to begin censoring. etc.Distribution Analysis Censoring indicators can be numbers or text. In the example above. while the higher one indicates an exact failure.

Weibull. 3-parameter lognormal. of the parametric distributions best fits your data. exponential. Fits one of eleven common parametric distributions to your data. Also draws survival. These plots help you determine which. lognormal. These help you assess the fit of the chosen distribution and view summary graphs of your data. 3-parameter Weibull. survival plot. Command Distribution ID Plot Right Censored Arbitrarily Censored Description Draws probability plots from your choice of eleven common distributions: smallest extreme value. and creates a Distribution ID Plot and Distribution Overview Plot. loglogistic. The commands include Parametric Distribution Analysis. if any. and cumulative failure probabilities. survival probabilities. These graphs are often used before the full analysis to help choose a distribution or view summary information. and hazard plot in separate regions on the same graph.Reliability and Survivial Analysis Here is the same data set structured both ways: Unstacked Data Drug A 20 30 43 51 57 82 85 89 Drug B 2 3 6 14 24 26 27 31 Stacked Data Drug 20 30 43 51 57 82 85 89 2 3 6 14 24 26 27 31 Group A A A A A A A A B B B B B B B B Note You cannot analyze more than one column of stacked data at a time. All rights reserved. Distribution Overview Plot Right Censored Arbitrarily Censored Parametric Distribution Analysis Right Censored Arbitrarily Censored Distribution ID Plot (Right Censoring) Stat > Reliability/Survival > Distribution Analysis (Right Censoring) > Distribution ID Plot Use Distribution ID Plot (Right Censoring) to determine which distribution best fits your data by comparing how closely the plot points lie to the best-fit lines of a probability plot. normal. 2parameter exponential. and 3-parameter loglogistic. and probability plots. Minitab also provides two goodness-of-fit measures to help you assess how the distribution fits your data: • • Anderson-Darling for the least squares and maximum likelihood estimation methods Pearson correlation coefficient for the least squares estimation method You can display up to 50 samples on each plot. probability density function. Draws a probability plot. logistic. 66 Copyright © 2003–2005 Minitab Inc. so the grouping indicators must be in one column. All the samples display on a single plot. cumulative failure. then uses that distribution to estimate percentiles. Distribution ID Plot Parametric distribution analysis commands You can use all parametric distribution analysis commands for both right-censored and arbitrarily-censored data. . Dialog box items Variables: Enter the columns of failure times. hazard. You can enter up to 50 columns (50 different samples). which performs the full analysis. Frequency columns (optional): Enter the columns of frequency data. with different colors and symbols.

your data are right-censored. logistic. If you have no censored values. enter the value you use to indicate censoring in Censoring value. Distribution 1: Check and choose one of eleven distributions: smallest extreme value. If all of the samples are stacked in one column. Specify: Choose to fit up to four distributions. Failure times are intermixed with censoring times. then enter the censoring columns in the box. If you do not enter a value. 3-parameter Weibull. • For data with censoring columns: Choose Use censoring columns.Distribution Analysis By variable: If all of the samples are stacked in one column. To make a distribution ID plot (uncensored/right-censored data) 1 2 3 4 5 Choose Stat > Reliability/Survival > Distribution Analysis (Right Censoring) > Distribution ID Plot. Distribution 4: Check and choose one of eleven distributions: smallest extreme value. exponential. Use all distributions: Choose to have Minitab fit all eleven distributions. You can enter up to 50 samples per analysis. exponential. All of the samples display on a single plot. lognormal. • When your data are multiply censored. you can skip steps 6 & 7. • Choose Specify to create up to four probability plots with the distributions of your choice. All rights reserved. Right censored data can be: • Singly censored − All of the test units run for the same amount of time. For time censored data: Choose Time censor at. Multiply censored − Test units are censored at different times. or 3parameter loglogistic. 2-parameter exponential. 3-parameter lognormal. Under certain conditions. lognormal. the second censoring column is paired with the second data column. Minitab can analyze systems with one cause of failure or multiple causes of failure. 3-parameter Weibull. 3-parameter lognormal. so you can compare the various functions between samples. logistic. or 3parameter loglogistic. 3-parameter Weibull. Do one of the following. logistic. and enter a column of grouping indicators in the box. normal. In Variables. see Multiple Failure Modes (Right Censored Data). Weibull. entering 500 says that any observation from 500 time units onward is considered censored. For systems that have more than one cause of failure. You can enter up to 50 columns (50 different samples). The first censoring column is paired with the first data column. Minitab allows you to draw conclusions based on that data. See Worksheet Structure for more information. • Copyright © 2003–2005 Minitab Inc. Weibull. For general information on life data and censoring. exponential. loglogistic. normal. Minitab estimates the functions independently for each sample. with different colors and symbols. you may have life data with no failures. 2-parameter exponential. 3-parameter lognormal. If you have frequency columns. enter the columns in Frequency columns. loglogistic. normal (default). see Distribution Analysis Data. Do one of the following: • Choose Use all distributions to create probability plots for all eleven distributions. and so on. or 3parameter loglogistic. Failed units are considered exact failures. Distribution 3: Check and choose one of eleven distributions: smallest extreme value. Minitab uses the lowest value in the censoring column. then click OK. or 3parameter loglogistic. 3-parameter Weibull. enter the columns of failure times. loglogistic. then enter a column of grouping indicators. where units go into service at different times. lognormal (default). logistic. 67 . Singly censored data are more common in controlled studies. you must have a column of censoring indicators. Note Occasionally. Distribution Analysis (Right Censored Data) If your data include exact failures or if test units do not fail before your study is over. If you like. 2-parameter exponential. loglogistic. check By variable. normal. exponential (default). See Drawing conclusions when you have few or no failures. Multiply censored data are more common in the field. Weibull. lognormal. Units surviving at the end of the study are considered censored data. unless you assume a common shape (Weibull) or scale (other distributions). Weibull (default). Note 6 7 Click Censor. check By variable. 2-parameter exponential. Distribution 2: Check and choose one of eleven distributions: smallest extreme value. 3-parameter lognormal. For example. then enter a failure time at which to begin censoring.

For example. Dialog box items Estimation Method Least Squares (failure time(X) on rank(Y)): Choose to estimate the distribution parameters using the least squares (XY) method. Time censor at: For time censored data. You can enter individual percents (0 < P < 100) or a column of percents. For example.Reliability and Survivial Analysis • For failure censored data: Choose Failure censor at. Maximum of tied points: Choose to plot the maximum cumulative percent of identical failure times. enter the value you use to indicate censoring in Censoring value. and so on. Minimum X scale: Enter a value for the minimum X scale. entering 150 says to make all (ordered) observations starting with the 150th observation censored. and all other observations uncensored. Dialog box items Censoring Options Use censoring columns: Choose Use censoring columns. and add your own title. You can also estimate percentiles for specified percents. choose Failure censor at. For example. then click OK. 8 If you like. You want to know − at given high temperatures − the time at which 1% of the engine windings fail. If you do not enter a value. type the desired text in this box. then enter a number of failures at which to begin censoring. then enter a number of failures at which to begin censoring. Engine windings may decompose at an unacceptable rate at high temperatures. Text values must be contained in double quotes. Failure censor at: For failure censored data. the second censoring column is paired with the second data column. choose Time censor at. . All rights reserved. Handle tied failure times by plotting: All points: Choose to plot the cumulative percents for each occurrence of a failure time. then enter the censoring columns in the box. Maximum Likelihood: Choose to estimate the distribution parameters using the maximum likelihood method. entering 500 says that any observation from 500 time units onward is considered censored. Minimum and Maximum X Scale Use default values: Choose to use the default values for the minimum and maximum X scale. entering 150 says to make all (ordered) observations starting with the 150th observation censored. use any of the available dialog box options. Title: To replace the default title. Use: Choose to enter your own values for the X scale minimum and maximum. specify the x-axis minimum and maximum. see Tools > Options > Individual Graphs > Probability Plots. and all other observations uncensored. Minitab uses the lowest value in the censoring column. Maximum X scale: Enter a value for the maximum X scale. Censor Stat > Reliability/Survival > Distribution Analysis (Right Censoring) > distribution analysis command > Censor Allows you to designate which observations are censored. which are estimated by fitting a regression line to the points in a probability plot. The first censoring column is paired with the first data column. Censoring value: If you like. which are estimated by maximizing the likelihood function. Example of a Distribution ID Plot for right-censored data Suppose you work for a company that manufactures engine windings for turbine assemblies. Average (median) of tied points: Choose to plot the median cumulative percent of identical failure times. Note To change the method for calculating probability plot points. You plan to get this information by using Parametric Distribution Analysis (Right 68 Copyright © 2003–2005 Minitab Inc. then enter a failure time at which to begin censoring. Distribution ID Plot (Right Censoring) − Options Stat > Reliability/Survival > Distribution Analysis (Right Censoring) > Distribution ID Plot > Options You can choose the method used to estimate the parameters. Show graphs of different variables or by levels: Choose to display the graphs overlaid on the same graph or on separate graphs. Estimate percentiles for these percents: Enter the additional percents for which you want to estimate percentiles.

5279 53. Choose Use censoring columns and enter Cens80 Cens100 in the box. and normal.8063 51.589 Correlation Coefficient 0. 1 2 3 4 5 Open the worksheet RELIABLE.98090 2.76414 27.9548 10.68093 1.44778 2. Leave the default distributions as Weibull.5365 62.63143 95% Normal CI Lower Upper 13.5539 69.Distribution Analysis Censoring).5156 40. Choose Stat > Reliability/Survival > Distribution Analysis (Right Censoring) > Distribution ID Plot. enter Temp80 Temp100. In the Minitab worksheet. you test 50 windings at 80° C.MTW. Choose Specify.7692 62.7777 56.0842 Distribution Weibull Lognormal Exponential Normal Mean 56.38997 7.2348 37.3558 50.606 67. which requires you to specify the distribution for your data.7339 21.7720 63.3722 33. Session window output Distribution ID Plot: Temp80. All rights reserved.2625 7.38602 2. Temp100 Results for variable: Temp80 Goodness-of-Fit Anderson-Darling (adj) 67.3846 56.9000 56.9618 34.5452 73.4791 69. In the first sample.7733 22.6859 23.73184 33.8663 56.5305 23.68499 27.9267 Copyright © 2003–2005 Minitab Inc.737540 15.656 71. in the second sample.76766 20.2497 54.46626 0.90747 3.2638 27. First you collect failure times for the engine windings at two temperatures.4812 28.6233 3.4020 5.21312 3.115720 4.1830 38.9307 2.1273 27.0893 28.542291 1.987 Distribution Weibull Lognormal Exponential Normal Table of Percentiles Standard Error 2.1804 62.41131 2.5033 50.0123 0.5140 2. 69 .71588 2.4005 51.2431 17. Click Censor. Click OK in each dialog box.11939 34.61498 0.8299 33.85714 2.00309 6.4955 62.982 * 0. lognormal.2361 5.5948 0. Some of the units drop out of the test for unrelated reasons.6179 61.0842 Distribution Weibull Lognormal Exponential Normal Weibull Lognormal Exponential Normal Weibull Lognormal Exponential Normal Weibull Lognormal Exponential Normal Percent 1 1 1 1 5 5 5 5 10 10 10 10 50 50 50 50 Percentile 17. you test 40 windings at 100° C. In Variables.48126 23.519 67.7692 33.97051 2.2635 51.986 0.9575 99.7394 3.9267 Table of MTTF Standard Error 2.8295 11.6552 40.7571 28.590591 3.6314 95% Normal CI Lower Upper 51. Distribution ID Plot − Right Censoring can help you choose that distribution. exponential. you use a column of censoring indicators to designate which times are actual failures (1) and which are censored units removed from the test before failure (0).

30292 7.0692 19. All rights reserved.9453 6.4961 14.3478 Table of MTTF Standard Error 3.98322 7.68409 22.21479 7.0816 34.17367 2.99440 4.637595 -21.0401 2.2166 Distribution Weibull Lognormal Exponential Normal Weibull Lognormal Exponential Normal Weibull Lognormal Exponential Normal Weibull Lognormal Exponential Normal Percent 1 1 1 1 5 5 5 5 10 10 10 10 50 50 50 50 Percentile 4.2681 47.70805 -4.1261 43.988 * 0.3137 31.17857 6.50848 4.3352 21.993 0.334673 0.9624 41.7510 0.5589 16.338 17.4474 4.25405 15.4401 34.5444 23.1357 64.23847 3.56053 3.1986 29.13270 10.62573 0.396 17. .3478 Distribution ID Plot for Temp80.78311 6.2997 63.50454 95% Normal CI Lower Upper 35.0759556 6.55752 3.09791 0.50454 95% Normal CI Lower Upper 2.8587 41.4790 48.97371 10.28281 2.796264 4.38694 14.84262 13.35756 5.44473 5.2166 Distribution Weibull Lognormal Exponential Normal Mean 42.960 Distribution Weibull Lognormal Exponential Normal Table of Percentiles Standard Error 1.8656 33.830 Correlation Coefficient 0.387650 5.9601 47.56130 9. Temp100 70 Copyright © 2003–2005 Minitab Inc.51233 9.461937 -9.07572 2.37705 0.88357 4.2969 37.39121 32.3453 49.4651 3.9733 48.35339 1.66562 0.281 19.4488 37.58784 6.90287 11.4790 15.76563 2.3297 39.5434 50.0222 3.0798 45.Reliability and Survivial Analysis Results for variable: Temp100 Goodness-of-Fit Anderson-Darling (adj) 17.6125 2.60714 1.

987 Distribution Weibull Lognormal Exponential Normal Copyright © 2003–2005 Minitab Inc.982 * 0. the Anderson-Darling values for the lognormal distribution are lower than the Anderson-Darling values for other distributions.606 67.589 Correlation Coefficient 0.656 71. Here. so the lognormal distribution would be a good choice when running the parametric distribution analysis. You can also compare the Anderson-Darling goodness-offit values to determine which distribution best fits the data. All rights reserved. 71 . Distribution ID Plot: Temp80 Goodness-of-Fit Anderson-Darling (adj) 67.Distribution Analysis Graph window output Interpreting the results The points fall approximately on the straight line on the lognormal probability plot. and 95% confidence intervals Probability plots for the chosen distributions For example. standard errors. The table of percentiles and MTTFs allow you to see how your conclusions may change with different distributions. thus supporting your conclusion that the lognormal distribution provides the best fit. their standard errors.986 0. Output − Distribution ID Plot The default output consists of: • • • • Goodness-of-fit statistics for the chosen distributions Table of percents and their percentiles. and 95% confidence intervals Table of MTTFs (mean time to failures). A smaller Anderson-Darling statistic means that the distribution provides a better fit.519 67.

and creates a Distribution ID Plot and Distribution Overview Plot.46626 0.9575 99.590591 3.4791 69.9548 10.3722 33. then uses that distribution to estimate percentiles.9267 Table of MTTF Standard Error 2. or a nonparametric overview plot. Fits one of eleven common parametric distributions to your data. and hazard plot in separate regions on the same graph.68499 27.63143 95% Normal CI Lower Upper 13. normal.7394 3.48126 23.5156 40.7339 21. 72 Copyright © 2003–2005 Minitab Inc.3558 50.71588 2. Command Distribution ID Plot Right Censored Arbitrarily Censored Description Draws probability plots from your choice of eleven common distributions: smallest extreme value.1830 38.00309 6.85714 2. Draws a probability plot. survival plot.5279 53.8295 11.76414 27. 3-parameter Weibull. Also draws survival.6233 3.73184 33.2431 17. Distribution Overview Plot Right Censored Arbitrarily Censored Parametric Distribution Analysis Right Censored Arbitrarily Censored Distribution Overview Plot (Right Censoring) Stat > Reliability/Survival > Distribution Analysis (Right Censoring) > Distribution Overview Plot Use Distribution Overview Plot to generate a layout of plots that allow you to view your life data in different ways on one graph.5033 50.2635 51.44778 2.737540 15.0893 28.2625 7.7692 33.0123 0. loglogistic.11939 34.90747 3.9618 34. All rights reserved.1273 27.6859 23. You can draw a parametric overview plot by selecting a distribution for your data.8063 51.0842 Distribution Weibull Lognormal Exponential Normal Weibull Lognormal Exponential Normal Weibull Lognormal Exponential Normal Weibull Lognormal Exponential Normal Percent 1 1 1 1 5 5 5 5 10 10 10 10 50 50 50 50 Percentile 17.7777 56.4812 28.6552 40.5452 73.5305 23.38602 2. probability density function. These help you assess the fit of the chosen distribution and view summary graphs of your data.1804 62.3846 56.7733 22. and probability plots.97051 2.6314 95% Normal CI Lower Upper 51. lognormal. which performs the full analysis. if any.8299 33. of the parametric distributions best fits your data.6179 61.0842 Distribution Weibull Lognormal Exponential Normal Mean 56.7692 62. 3-parameter lognormal.2348 37.41131 2.9000 56.2497 54. and 3-parameter loglogistic. The commands include Parametric Distribution Analysis. survival probabilities.542291 1. hazard.61498 0.9267 Distribution ID Plot for Temp80 Overview Plot Parametric distribution analysis commands You can use all parametric distribution analysis commands for both right-censored and arbitrarily-censored data.4005 51. These plots help you determine which.5365 62.7571 28.68093 1.5539 69. cumulative failure.4020 5.98090 2. and cumulative failure probabilities.2361 5.9307 2.4955 62. Weibull.5140 2. logistic.38997 7.76766 20.8663 56.Reliability and Survivial Analysis Table of Percentiles Standard Error 2.115720 4. 2parameter exponential. exponential.2638 27.21312 3. . These graphs are often used before the full analysis to help choose a distribution or view summary information.5948 0.7720 63.

you can skip steps 6 & 7. see Distribution Analysis Data. then click OK. 3-parameter lognormal. See Worksheet Structure for more information. Minitab estimates the functions independently for each sample. Units surviving at the end of the study are considered censored data. For systems that have more than one cause of failure. Singly censored data are more common in controlled studies. Weibull (default). enter the columns of failure times. and enter a column of grouping indicators in the box Parametric analysis: Choose to perform a parametric distribution analysis. 73 . a probability density function. which helps you compare their various functions. see one of the Distribution Analysis Commands.Distribution Analysis The parametric display includes a probability plot (for a selected distribution). with different colors and symbols. Note 6 7 Click Censor. By variable: If all of the samples are stacked in one column. For general information on life data and censoring. 2-parameter exponential. you may have life data with no failures. All of the samples display on a single plot. Minitab displays a Turnbull survival plot or an Actuarial survival plot and hazard plot. Do one of the following. From Distribution. Failure times are intermixed with censoring times. and enter a column of grouping indicators in the box. Minitab estimates the functions independently for each sample. you must have a column of censoring indicators. All rights reserved. Right censored data can be: • Singly censored − All of the test units run for the same amount of time. logistic. Minitab allows you to draw conclusions based on that data. These functions are all typical ways of describing the distribution of failure time data. Copyright © 2003–2005 Minitab Inc. Choose to draw a parametric or nonparametric plot: • Parametric plot − Choose Parametric analysis. Note Occasionally. To make a distribution overview plot (uncensored/right-censored) 1 2 3 4 5 Choose Stat > Reliability/Survival > Distribution Analysis (Right Censoring) > Distribution Overview Plot. normal. a survival (or reliability) plot. Nonparametric analysis: Choose to perform a nonparametric distribution analysis. check By variable. your data are right-censored. Dialog box items Variables: Enter the columns of failure times. In Variables. If all of the samples are stacked in one column. see Multiple Failure Modes (Right Censored Data). The nonparametric display depends on the type of data: if you have right-censored data Minitab displays a Kaplan-Meier survival plot and a hazard plot or an Actuarial survival plot and hazard plot. Multiply censored − Test units are censored at different times. or 3-parameter loglogistic. • When your data are multiply censored. and if you have arbitrarily-censored data. Minitab can analyze systems with one cause of failure or multiple causes of failure. Under certain conditions. You can enter up to 50 columns (50 different samples). lognormal. To draw these plots with more information. Frequency columns (optional): Enter the columns of frequency data. and a hazard plot. unless you assume a common shape (Weibull) or scale (other distributions). Multiply censored data are more common in the field. Failed units are considered exact failures. exponential. where units go into service at different times. so you can compare the various functions between samples. Distribution Analysis (Right Censored Data) If your data include exact failures or if test units do not fail before your study is over. • Nonparametric plot − Choose Nonparametric analysis. All of the samples display on a single plot. loglogistic. enter the columns in Frequency columns. choose to plot one of the eleven available distributions. check By variable. in different colors and symbols. 3-parameter Weibull. You can enter up to 50 samples per analysis. Distribution: Choose one of eleven distributions: smallest extreme value. You can enter up to 50 columns (50 different samples). If you have no censored values. See Drawing conclusions when you have few or no failures. If you have frequency columns.

then enter a number of failures at which to begin censoring. and leave all other observations uncensored. Minitab uses the lowest value in the censoring column. If you like. All rights reserved. Dialog box items Censoring Options Use censoring columns: Choose Use censoring columns.Reliability and Survivial Analysis • For data with censoring columns: Choose Use censoring columns. Average (median) of tied points: Choose to plot the median cumulative percent of identical failure times. type the desired text in this box. The first censoring column is paired with the first data column. Dialog box items Estimation Method Least Squares (failure time(X) on rank(Y)): Choose to estimate the distribution parameters using the least squares (XY) method. entering 500 says that any observation from 500 time units onward is considered censored. Minimum and Maximum X Scale Use default values: Choose to use the default values for the minimum and maximum X scale. the second censoring column is paired with the second data column. entering 500 says that any observation from 500 time units onward is considered censored. If you do not enter a value. For example. Handle tied failure times by plotting: All points: Choose to plot the cumulative percents for each occurrence of a failure time. Maximum X scale: Enter a value for the maximum X scale. Maximum of tied points: Choose to plot the maximum cumulative percent of identical failure times. 74 Copyright © 2003–2005 Minitab Inc. Minimum X scale: Enter a value for the minimum X scale. then click OK. Censor Stat > Reliability/Survival > Distribution Analysis (Right Censoring) > distribution analysis command > Censor Allows you to designate which observations are censored. enter the value you use to indicate censoring in Censoring value. specify the x-axis minimum and maximum. Note To change the method for calculating probability plot points. For example. Distribution Overview Plot (Right Censoring) − Options Stat > Reliability/Survival > Distribution Analysis (Right Censoring) > Distribution Overview Plot > Options You can choose the method used to estimate the parameters. and all other observations uncensored. and so on. The first censoring column is paired with the first data column. enter the value you use to indicate censoring in Censoring value. choose Time censor at. For failure censored data: Choose Failure censor at. Maximum Likelihood: Choose to estimate the distribution parameters using the maximum likelihood method. then enter a failure time at which to begin censoring. Show graphs of different variables or by levels: Choose to display the graphs overlaid on the same graph or on separate graphs. and so on. then enter a failure time at which to begin censoring. Title: To replace the default title. Failure censor at: For failure censored data. Censoring value: If you like. entering 150 says to censor all (ordered) observations from the 150th observed failure on. Use: Choose to enter your own values for the X scale minimum and maximum. which are estimated by maximizing the likelihood function. see Tools > Options > Individual Graphs > Probability Plots. and add your own title. which are estimated by fitting a regression line to the points in a probability plot. choose Failure censor at. then enter the censoring columns in the box. Text values must be contained in double quotes. For example. then enter a number of failures at which to begin censoring. For example. If you do not enter a value. Minitab uses the lowest value in the censoring column. use any of the available dialog box options. then enter the censoring columns in the box. You can also define how ties are to be handled. 8 If you like. entering 150 says to make all (ordered) observations starting with the 150th observation censored. • • For time censored data: Choose Time censor at. the second censoring column is paired with the second data column. . Time censor at: For time censored data.

choose Lognormal. you test 40 windings at 100° C. in the second sample. First you collect data for times to failure for the engine windings at two temperatures. In the Minitab worksheet. enter Temp80 and Temp100. These units are considered to be right censored because their failures were not due to the cause of interest. at given high temperatures. you test 50 windings at 80° C. you use a column of censoring indicators to designate which times are actual failures (1) and which are censored units removed from the test before failure (0). 1 2 3 4 5 Open the worksheet RELIABLE. In the first sample. You want to know.MTW. Choose Use censoring columns and enter Cens80 and Cens100.656 Correlation Coefficient 0. Some of the units drop out of the test due to failures from other causes. 75 . Temp100 Copyright © 2003–2005 Minitab Inc. Choose Stat > Reliability/Survival > Distribution Analysis (Right Censoring) > Distribution Overview Plot. Click Censor. In Distribution. Engine windings may decompose at an unacceptable rate at high temperatures.Distribution Analysis Example of a Distribution Overview Plot with right-censored data Suppose you work for a company that manufactures engine windings for turbine assemblies.982 Distribution Lognormal Results for variable: Temp100 Goodness-of-Fit Anderson-Darling (adj) 17. Session window output Distribution Overview Plot: Temp80.281 Correlation Coefficient 0. In Variables. You plan to get this information by using Parametric Distribution Analysis (Right Censoring). All rights reserved. Temp100 Results for variable: Temp80 Goodness-of-Fit Anderson-Darling (adj) 67. but you first want to have a quick look at your data from different perspectives. Click OK in each dialog box. at what time do 1% of the engine windings fail.988 Distribution Lognormal Distribution Overview Plot for Temp80.

With these plots. you get: • • • • • • Goodness-of-fit statistics for the chosen distribution Probability plot.Nonparametric hazard plot based on the empirical hazard function For right-censored data with Actuarial method . or f(y). Output − Distribution Overview Plot The distribution overview plot display differs depending on whether you select the parametric or nonparametric display. Parametric hazard plot. Probability density function.Actuarial survival plot . failure time.Actuarial survival plot . failure time. which displays the hazard function or instantaneous failure rate. which displays estimates of the cumulative distribution function F(y) vs. failure time.Turnbull survival plot For arbitrarily-censored data with Actuarial method . you get: • • • 76 Copyright © 2003–2005 Minitab Inc. .Reliability and Survivial Analysis Graph window output Interpreting the results These four plots describe the failure rate of engine windings at two different temperatures. For right-censored data with Kaplan-Meier method . All rights reserved. which displays the curve that describes the distribution of your data. Parametric survival (or reliability) plot. When you select a parametric display. which displays the survival (or reliability) function 1-F(y) vs. f(y)/(1-F(y)) vs.Nonparametric hazard plot based on the empirical hazard function For arbitrarily-censored data with Turnbull method . you can determine how much more likely it is that engine windings will fail when running at 100° C as opposed to 80° C.Kaplan-Meier survival plot .Nonparametric hazard plot based on the empirical hazard function When you select a nonparametric display.

check By variable. Command Distribution ID Plot Right Censored Arbitrarily Censored Description Draws probability plots from your choice of eleven common distributions: smallest extreme value. You can fit one of eleven common distributions to your data. and cumulative failure probabilities. evaluate the appropriateness of the distribution. and empirical hazard function change values only at exact failure times. estimate percentiles. 2-parameter exponential. loglogistic. see Distribution ID Plot (Right Censoring) which draws four probability plots in separate regions on the same graph. lognormal. normal. These help you assess the fit of the chosen distribution and view summary graphs of your data. exponential. Weibull (default). Frequency columns (optional): Enter the columns of frequency data. and 3-parameter loglogistic. 2parameter exponential. Distribution Overview Plot Right Censored Arbitrarily Censored Parametric Distribution Analysis Right Censored Arbitrarily Censored Parametric Distribution Analysis (Right Censoring) Stat > Reliability/Survival > Distribution Analysis (Right Censoring) > Parametric Distribution Analysis Use Parametric Distribution Analysis − Right Censoring when your data are right censored or include actual failure times. 3-parameter Weibull. Use the probability plot to see whether the distribution fits your data. logistic. 77 . probability density function. survival plot. These plots help you determine which. You can enter up to 50 columns (50 different samples). For example. if any.Distribution Analysis The Kaplan-Meier survival estimates. Distribution Overview Plot: Temp80 Goodness-of-Fit Anderson-Darling (adj) 67. exponential. so the nonparametric survival and hazard curves are step functions. and draw survival. which performs the full analysis. normal. and creates a Distribution ID Plot and Distribution Overview Plot. Copyright © 2003–2005 Minitab Inc. survival probabilities. 3-parameter lognormal. hazard. 3parameter Weibull. see Distribution Overview Plot (Right Censoring). Weibull. use one of the nonparametric distribution analyses. These graphs are often used before the full analysis to help choose a distribution or view summary information. hazard. and enter a column of grouping indicators in the box Assumed distribution: Choose one of eleven common distributions: smallest extreme value. then uses that distribution to estimate percentiles. To view your data in different ways.986 Distribution Weibull Distribution Overview Plot for Temp80 Parametric Distribution Analysis Parametric distribution analysis commands You can use all parametric distribution analysis commands for both right-censored and arbitrarily-censored data. Turnbull survival estimates. Also draws survival. and probability plots. cumulative failure. The commands include Parametric Distribution Analysis. All rights reserved. Dialog box items Variables: Enter the columns of failure times. Draws a probability plot. and probability plots. 3-parameter lognormal. Fits one of eleven common parametric distributions to your data. By variable: If all of the samples are stacked in one column. Parametric survival and hazard estimates are based on a fitted distribution and the curve will therefore be smooth. and hazard plot in separate regions on the same graph. To compare the fits of different distributions. cumulative failure. survival probabilities. or 3-parameter loglogistic. If no parametric distribution fits your data. and cumulative failure probabilities. of the parametric distributions best fits your data. lognormal. loglogistic.606 Correlation Coefficient 0. logistic.

by default Minitab uses the lowest value in the censoring column. For failure censored data: Choose Failure censor at. Do one of the following. For example. You can enter up to 50 samples per analysis. Failure times are intermixed with censoring times. unless you assume a common shape (Weibull) or scale (other distributions). such as hours. and so on. Minitab allows you to draw conclusions based on that data. If you have no censored values. See Worksheet Structure for more information. Under certain conditions. γ . you must have a column of censoring indicators. with different colors and symbols. you may have life data with no failures. Note 5 6 Click Censor. All rights reserved. where units go into service at different times. • For data with censoring columns: Choose Use censoring columns. Note Occasionally. For general information on life data and censoring.Reliability and Survivial Analysis Distribution Analysis (Right Censored Data) If your data include exact failures or if test units do not fail before your study is over. To do a parametric distribution analysis (uncensored/right censored data) 1 2 3 4 Choose Stat > Reliability/Survival > Distribution Analysis (Right Censoring) > Parametric Distribution Analysis. then enter a failure time at which to begin censoring. check By variable. Choosing a Distribution with a Threshold Parameter The threshold parameter. • When your data are multiply censored. enter the value you used to indicate censoring in Censoring value. You can enter up to 50 columns (50 different samples). the second censoring column is paired with the second data column. then enter the censoring columns in the box. Failed units are considered exact failures. so you can compare the various functions between samples. or cycles. The threshold parameter locates the distribution along the time scale and has the same units as time. Minitab estimates the functions independently for each sample. All of the samples display on a single plot. Multiply censored − Test units are censored at different times. your data are right-censored. If you have frequency columns. and leave all other observations uncensored. Minitab can analyze systems with one cause of failure or multiple causes of failure. and enter a column of grouping indicators in the box. enter the columns in Frequency columns. see Multiple Failure Modes (Right Censored Data). If all of the samples are stacked in one column. . entering 500 says that any observation from 500 time units onward is considered censored. 78 Copyright © 2003–2005 Minitab Inc. see Distribution Analysis Data. miles. provides an estimate of the earliest time a failure may occur. enter the columns of failure times. then click OK. Right censored data can be: • Singly censored − All of the test units run for the same amount of time. For systems that have more than one cause of failure. 7 If you like. use any dialog box options. • • For time censored data: Choose Time censor at. then enter a number of failures at which to begin censoring. In Variables. Multiply censored data are more common in the field. The first censoring column is paired with the first data column. If you do not enter a value. then click OK. Units surviving at the end of the study are considered censored data. If you like. See Drawing conclusions when you have few or no failures. you can skip steps 5 & 6. Singly censored data are more common in controlled studies. entering 150 says to censor all (ordered) observations from the 150th observation on. For example.

Copyright © 2003–2005 Minitab Inc. 3-parameter lognormal. In this example. A negative γ indicates that failures have occurred prior to the beginning of a test. The parametric distribution analysis commands automatically display a table of percentiles in the Session window. Choose a distribution with a threshold parameter (3-parameter Weibull. 2-parameter exponential. Percentiles By what time do half of the engine windings fail? How long until 10% of the blenders stop working? You are looking for percentiles. When γ > 0. or 3-parameter loglogistic) when you want to estimate the earliest time-to-failure. the distribution starts to the right of the origin. 40. 20. All rights reserved. The period from 0 to γ is the failure free operating period. 79 . The two probability plots show the same data fit to a Weibull and a 3-parameter Weibull distribution. The Weibull does not account for the threshold parameter and displays as a curve on probability paper. the distribution starts to the left of the origin. 30. 80. Note The threshold parameter is assumed fixed when calculating confidence intervals with the 3-parameter lognormal and 2-parameter exponential distributions. we entered failure times (in months) for engine windings. Minitab displays the percentiles 1-10.Distribution Analysis • • • When γ = 0. The 3-parameter Weibull adjusts for γ and the points appear straighter. 50. When γ < 0. the distribution starts at the origin. 60. 70. By default. and 90-99.

1% of the windings failed. choose Time censor at. Parametric Distribution Analysis − Failure Mode Stat > Reliability/Survival > Distribution Analysis (Right Censoring) or Distribution Analysis (Arbitrary Censoring) > Parametric Distribution Analysis > FMode Use to estimate the overall reliability of your system when multiple causes of failure exist in order to investigate the reliability of the individual failure modes. The first censoring column is paired with the first data column. Time censor at: For time censored data. then enter a number of failures at which to begin censoring. enter the value you use to indicate censoring in Censoring value. you can specify a different confidence level for all confidence intervals.5009 27. entering 500 says that any observation from 500 time units onward is considered censored. To represent censored observations in the failure mode column. 80 Copyright © 2003–2005 Minitab Inc. Here.6767 24. All rights reserved.76% of the engine windings last past 70 months. then enter a failure time at which to begin censoring. The table also includes standard errors and approximate 95. for example.0% Normal CI Time Probability Lower 70.4489 Standard Error 2. which are estimates of the proportion of units that survive past a given time.78453 3. Minitab uses the lowest value in the censoring column.5222 As shown in the table above. The values in the Percentile column are estimates of the times at which the corresponding percent of the units failed. Censoring value: If you like.7601 12. Dialog box items Use failure mode columns: Enter the columns containing the failure modes.0% confidence intervals for each percentile. Use failure modes: Enter the failure modes to include in the analysis.48898 3. 40. choose Failure censor at. at about 10 months (Percentile).2894 Upper 0. Survival Probabilities − Parametric Distribution Analysis What is the probability of an engine winding running past a given time? How likely is it that a cancer patient will live five years after receiving a certain drug? You are looking for survival probabilities.86263 17. use an asterisk (*) for numeric columns or a space for text columns. If you do not enter a value.3193 8. the second censoring column is paired with the second data column. In the Estimate subdialog box. entering 150 says to make all (ordered) observations starting with the 150th observation censored. When you request survival probabilities in the Estimate subdialog box.23157 3.2270 As shown in the first row of the table.2590 18. This option is available for both right-censored and arbitrarily censored data.Reliability and Survivial Analysis Table of Percentiles Percent 1 2 3 4 Percentile 10.4076 0. For example. and so on.6193 16. and all other observations uncensored. You can also request percentiles to be added the default table.0000 0.55426 21. we requested a survival probability for engine windings running at 70 months: Table of Survival Probabilities 95.0765 13. Text values must be contained in double quotes. Eliminate failure modes: Enter the failure modes to exclude from the analysis.0% Normal CI Lower Upper 5. then enter the censoring columns in the box. Dialog box items Censoring Options Use censoring columns: Choose Use censoring columns. Failure censor at: For failure censored data. the parametric distribution analysis commands display them in the Session window. Failure Mode Options Use all failure modes: Choose to include all failure modes in the analysis. For example.6834 10. .66352 95. Censor Stat > Reliability/Survival > Distribution Analysis (Right Censoring) > distribution analysis command > Censor Allows you to designate which observations are censored.

enter Failure.9579 Parameter Location Scale Estimate 124. enter 52. so you can redesign it to optimize overall system reliability.Distribution Analysis Create right censored observations using (Only available with arbitrary censoring. Click Results. and Distribution ID Plot (Right Censoring) can help you choose appropriate distributions. Click FMode.1805 39. In Use failure mode columns. You plan to get this information using Parametric Distribution Analysis (Right Censoring).153 29. You can specify a distribution for each failure mode. Session window output Distribution Analysis: Weeks Variable: Weeks Failure Mode: Failure = Meter Censoring Information Uncensored value Right censored value Count 20 60 Estimation Method: Least Squares (failure time(X) on rank(Y)) Distribution: Logistic Parameter Estimates Standard Error 9. Click OK in each dialog box. Right endpoint of intervals: Choose if the experiment continues until the right endpoint when a failure in the interval occurs. enter Weeks.0% Normal CI Lower Upper 104.843 Correlation Coefficient = 0.) Use to determine how Minitab will create right-censored observations for other failure modes when data are interval censored. Example of a parametric distribution analysis with multiple failure modes You are responsible for improving the overall reliability of pressure sensing circuits.811 Goodness-of-Fit Anderson-Darling (adjusted) = 2.97694 4. Choose Stat > Reliability/Survival > Distribution Analysis (Right Censoring) > Parametric Distribution Analysis In Variables. leading to system failure: sensor. Change Distribution for Levels Use only if distribution is different from that selected in the main dialog box. 81 . 1 2 3 4 5 6 7 8 9 Open the worksheet CIRCUIT.985 Copyright © 2003–2005 Minitab Inc.71076 95.MTW.598 143. Midpoint of intervals: Choose if a failure for any one failure mode causes the experiment to end. 10 Check Display analyses for individual failure modes according to display of results. transmitter. then choose the corresponding distribution for each failure mode. and meter.0917 Log-Likelihood = -120. Level: Enter failure mode. In Estimate probabilities for these times (values). All rights reserved.707 21. Under Change Distribution for Levels. Click Estimate. You want to determine which component fails most frequently. Click OK. Click OK. type "Meter" in Level and choose Logistic. Three main components could fail.

598 143.461 195.962510 Time 52 Probability 0.5036 112.574 312.235 236.501 226.373 257.4690 21.922741 Distribution Analysis: Weeks Variable: Weeks Failure Mode: Failure = Sensor Censoring Information Uncensored value Right censored value Count 30 50 Estimation Method: Least Squares (failure time(X) on rank(Y)) Distribution: Weibull 82 Copyright © 2003–2005 Minitab Inc.8232 99.200 209.88319 9.9211 Table of Percentiles Standard Error 29.7807 109.5679 27.0% Normal CI Lower Upper 104.88367 13.1807 39.2312 16.833 Table of Survival Probabilities 95.153 52.608 225.525 203.3624 14.0028 -5.847471 0.4672 13.685 269.3656 18.949 148.092 Percent 0.97694 8.1060 48.54438 9.4941 44.153 92.074 191.221 285.2138 130.969 136.97694 10.1924 156.611 82.8370 18.1957 68.0104 71.0% Normal CI Lower Upper 0.501 104.1474 35.4172 72.1805 10.1 1 2 3 4 5 6 7 8 9 10 20 30 40 50 60 70 80 90 91 92 93 94 95 96 97 98 99 Percentile -76.1366 43.6836 95.153 135.113 63.417 157.357 124. All rights reserved.0% Normal CI Lower Upper -134.339 154.6794 11.0272 31.2128 17.9454 4.598 143.947 46.9331 23.604 130.7965 Mean(MTTF) Standard Deviation Median First Quartile(Q1) Third Quartile(Q3) Interquartile Range(IQR) Estimate 124.2951 -47.5098 78.636 172.52708 10.201 162.258 -19.3301 14.3268 63.404 204.92846 58.0162 26.5767 9.1653 -21.6614 20.802 164.279 237.5018 80.7666 124.7765 -9. .7929 31.5383 87.626 192.873 189.6978 38.9856 12.812 216.3506 95.4663 22.0350 101.25702 9.707 74.8453 60.89101 51.4756 104.0928 116.6581 12.573 165.3278 19.9640 19.2318 83.97694 8.07580 8.3632 17.0542 75.914 94.280 181.480 124.7545 13.598 143.105 156.9016 53.8856 10.535 258.721 169.043 222.9619 66.8407 12.6538 20.930 174.693 249.420 159.Reliability and Survivial Analysis Characteristics of Distribution Standard Error 9.707 38.2195 28.209 231.205 199.483 188.1993 11.679 242.7524 24.707 114.1008 56.681 180.2485 10.

6444 99.4188 42.99962 7.7409 98.5176 76.203 Table of Survival Probabilities 95.347 172.2279 111.9431 89.72977 9.17657 18.782 134.1759 Log-Likelihood = -184.4436 Mean(MTTF) Standard Deviation Median First Quartile(Q1) Third Quartile(Q3) Interquartile Range(IQR) Estimate 88.468 191.98116 3.27683 7.8751 96.60396 4.0923 106.2421 100.248 130.237 138.4832 35.54882 95.5337 68.84987 6.8477 74.599 48.5291 34.3708 Table of Percentiles Standard Error 1.164 106.000 Parameter Shape Scale Estimate 2.812 95.973 Percent 0.2891 38.4549 67.25677 5.25215 8.225 136.0910 31.1016 30.0255 40.442 123.5211 48.3477 85.7705 79.08268 5.213 144.845809 Copyright © 2003–2005 Minitab Inc.107 180.117 147.2581 43.1 1 2 3 4 5 6 7 8 9 10 20 30 40 50 60 70 80 90 91 92 93 94 95 96 97 98 99 Percentile 8.76775 99.3477 53.69945 95.72493 3.3079 48.6113 111. 83 .99719 3.3791 26.91618 8.4159 32.1467 13.984 Characteristics of Distribution Standard Error 5.391 154.252 123.52483 4.053 136.318 153.6227 54.12855 7.08543 13.49793 5.34505 3.4694 17.67210 4.17628 4.8185 24.8043 86.2260 33.4217 43.309 127.0% Normal CI Lower Upper 0.19298 6.7409 98.301 156.0805 30.9840 57.45040 7.647 156.36804 95.755073 0.2730 34.905001 Time 52 Probability 0.3345 77.99019 4.7388 21.567 121.9152 32.39714 5.4842 86.65842 7.3316 50.1880 118.640 140.64198 5.49793 5.5241 28.0252 52.214 148.31797 4.425 Correlation Coefficient = 0.187 150.0% Normal CI Lower Upper 78.6828 68.055 117.43125 4.690 173.0968 76.2183 28.5651 46.2646 37.234031 5.73159 5.0% Normal CI Lower Upper 5. All rights reserved.555 167.26664 88.9114 36.485 133.054 162.609 141.671 125.18047 5.8751 63.5566 58.2002 55.6467 107.5844 26.153 163.37997 5.20180 2.0919 39.Distribution Analysis Parameter Estimates Standard Error 0.0% Normal CI Lower Upper 2.235 Goodness-of-Fit Anderson-Darling (adjusted) = 91.095 146.27552 3.994 42.424 151.220 130.230 159.1087 24.4409 40.88597 6.

1312 20.550 39.6753 20.676 319.6300 100.292 96.0% Normal CI Lower Upper 1.226 214.0147 18.0266 39.5584 30.17112 7.72088 103.019 252.6474 94.0909 57.64307 4.084 149.229 165.4391 85.671 145.181461 9.6738 98.4292 20.383 182.733 251.1735 8.1494 33.8738 70.1 1 2 3 4 5 6 7 8 9 10 20 30 40 50 60 70 80 90 91 92 93 94 95 96 97 98 99 Percentile 1.4297 24.3169 56.0751 83.6138 16.Reliability and Survivial Analysis Distribution Analysis: Weeks Variable: Weeks Failure Mode: Failure = Transmitter Censoring Information Uncensored value Right censored value Count 30 50 Estimation Method: Least Squares (failure time(X) on rank(Y)) Distribution: Weibull Parameter Estimates Standard Error 0.917 Percent 0.7163 19.81901 3.6300 100.7242 13.2534 Mean(MTTF) Standard Deviation Median First Quartile(Q1) Third Quartile(Q3) Interquartile Range(IQR) Estimate 92.75168 95.39957 2.9521 8.0159 26.4098 16.14758 10.4412 26.14057 7.9741 12. All rights reserved.397 153.16923 2.0% Normal CI Lower Upper 77.100 172.792 42.987 Characteristics of Distribution Standard Error 8.272 182.488 Goodness-of-Fit Anderson-Darling (adjusted) = 182.773 188.475 84 Copyright © 2003–2005 Minitab Inc.9509 137.327 136.824159 2.8130 13.43608 3.185 197.857 195.11596 86.516 168.3922 15.875 204.537 223.8855 95.10256 4.244 Parameter Shape Scale Estimate 1.5514 22.15160 11.049 241.84373 6.84373 8.1633 17.6738 50.5000 71.140 138.30450 4.4909 69.175 74.2114 27.9072 71.47457 7.528 177.78821 5. .788576 4.237 163.7783 17.107 59.8892 18.177 287.0606 35.9775 123.3038 55.5464 24.549 Correlation Coefficient = 0.8233 105.94300 12.4064 115.2711 83.0% Normal CI Lower Upper 0.186 114.535 Log-Likelihood = -187.86991 4.87034 7.2113 24.2130 12.10925 7.1951 125.25814 3.5641 37.48302 4.5085 32.1978 69.747 172.550 82.40640 17.150 149.8139 22.568 216.1392 18.5614 117.3062 56.965 209.642 228.9566 6.2905 45.4770 63.59506 3.20506 95.801 141.639 159.81472 10.976 231.9798 Table of Percentiles Standard Error 0.5175 17.2546 28.8970 30.7618 10.51657 21.6011 109.375 267.0006 43.653 203.79410 6.

296 111.8862 32.0704 93.74528 21. All rights reserved.8054 67.5891 111.1 1 2 3 4 5 6 7 8 9 10 20 30 40 50 60 70 80 90 91 92 93 94 95 96 97 98 99 Percentile -76.5031 65.4556 41.032 99.4550 20.605 124.518 105.815503 Time 52 Probability 0.5881 14. Weibull.0% Normal CI Lower Upper 0.6733 77.614 104.1079 4.6076 58.7417 114.180 Percent 0.736598 Distribution Analysis: Weeks Variable: Weeks Failure Mode: Failure = Meter.623 92.8527 39.9356 19.35150 17.870 133.5580 10.6968 0. 85 .664499 Time 52 Probability 0.3074 17.8017 107.293470 15.1316 116.77289 19.506 118.72617 10.1932 42.5579 50.2951 -47.838 119.3122 99.9494 12.500 Table of Survival Probabilities 95.0459 25.27227 -16.136 121.0254 82.18134 7.4591 6.52708 4.9700 85.8636 34.669 142.023 132.639 101.7293 57.4906 26.1511 74.0% Normal CI Lower Upper -134.157 114.3531 12.6745 49.456 103.0% Normal CI Lower Upper 0.6042 8.1127 90.698 96.5768 14.632389 0.993 123.4731 23.576 94.2195 9. Transmitter Censoring Information Uncensored value Count 80 Estimation Method: Least Squares (failure time(X) on rank(Y)) Distribution: Logistic.Distribution Analysis Table of Survival Probabilities 95.9792 57.5548 101.411 113.8514 28.6213 109.5404 16.7766 -9.0401 25.110 108.4426 73.480067 0.258 -19.3707 12.778 108. Weibull Table of Percentiles 95. Sensor.9891 66.832 127.0185 13.574887 Copyright © 2003–2005 Minitab Inc.7026 49.

and transmitters respectively survive past 52 weeks. Maximum Likelihood: Choose to estimate the distribution parameters using the maximum likelihood method. 0. see Least squares estimates versus maximum likelihood estimates. sensors. You can also enter a confidence level that Minitab will use for all confidence intervals.922741. That is. Parametric Distribution Analysis − Estimate Stat > Reliability/Survival > Distribution Analysis (Right Censoring) or Distribution Analysis (Arbitrary Censoring) > Parametric Distribution Analysis > Estimate Choose the estimation used to estimate percentiles and survival probabilities for specified values. .Reliability and Survivial Analysis Graph window output Interpreting the results The overall 52-week circuit reliability is 0. Set threshold at: Enter a value for the threshold parameter for all the variables. That is. and 74% of the meters. Minitab estimates threshold parameters if you do not provide values to use. 86 Copyright © 2003–2005 Minitab Inc. 57% of the circuits survive past 52 weeks.574887. 84%. and transmitter. 0. To improve the overall reliability.736598. you may need to improve both the sensor and the transmitter. which maximizes the likelihood function. sensor.845809. Estimate percentiles for these additional percents: Enter the additional percents for which you want to estimate percentiles.664499. 0. You can enter individual percents (0 < P < 100) or a column of percents. Assume common shape (slope−Weibull) or scale (1/slope−other dists): Check to estimate the parameters while assuming a common shape or scale parameter. All rights reserved. or enter a list of values for the shape or scale that is equal in length to the number of variables. 92%. or enter a list of values equal to the number of response variables. which fits a regression line to the points in a probability plot. Bayes analysis Set shape (slope−Weibull) or scale (1/slope−other dists) at: Enter a value for the shape or scale parameter for all the response variables. For more information on selecting estimation methods. Dialog box items Estimation Method Least Squares (failure time(X) on rank(Y)): Choose to estimate the distribution parameters using the least squares (XY) method. You are 95% confident that the true reliability is between 0. The 52-week survival probability for each of the components is: meter.480067 and 0.

click Estimate. For distributions with threshold parameter: • To estimate the scale and threshold parameters while keeping the shape fixed. enter the additional percents for which you want to estimate percentiles. 87 . • To estimate the location parameter while keeping the scale fixed (other distributions). In Estimate percentiles for these additional percents. under Bayes Analysis: Copyright © 2003–2005 Minitab Inc. Under Estimation Method. To draw conclusions when you have few or no failures 1 3 2 In the main dialog box. enter one value to be used for all samples. click Estimate. click Estimate. Click OK. or a series of values. Estimate survival probabilities: Choose to estimate survival probabilities. click Estimate. enter one value to be used for all samples. In Estimate probabilities for these times (values). To estimate one parameter while keeping the other parameter fixed 1 2 In the main dialog box. enter one value to be used for all samples. In Confidence level. To choose the method for estimating parameters 1 2 3 In the main dialog box. or a column of times for which you want to calculate survival probabilities. or a series of values. Depending on your distribution. Do one of the following: For all distributions: • To estimate the scale parameter while keeping the shape fixed (Weibull). or a series of values. see Starting estimates for more information. 3 To request parametric survival probabilities 1 2 3 In the main dialog box. click Estimate. enter one or more times. To change confidence levels 1 2 3 In the main dialog box. Click OK. Under Estimation Method. In Set shape (slope−Weibull) or scale (1/slope−other dists) at. see Starting estimates for more information. In Set shape (slope−Weibull) or scale (1/slope−other dists) at. Click OK.0%. Confidence level: Enter a confidence level for all of the confidence intervals. All rights reserved. enter one value to be used for all samples. click Estimate. You can enter individual percents (0 < P < 100) or a column of percents. Confidence intervals: Choose to use two-sided confidence intervals (the default) or just an upper or lower confidence interval. enter a value. In Set shape (slope−Weibull) or scale (1/slope−other dists) at. see Starting estimates for more information. To request additional percentiles 1 2 3 In the main dialog box. Estimate cumulative failure probabilities: Choose to estimate cumulative failure probabilities.Distribution Analysis Estimate probabilities for these times (values): Enter one or more times or a column of times for which you want to calculate survival probabilities or cumulative failure probabilities. Click OK. The default is 95. or a series of values. Click OK. choose Least Squares (the default) or Maximum Likelihood. choose Maximum Likelihood. In Set threshold at. • To estimate the scale and shape parameters while keeping the threshold fixed.

and threshold. enter shape and scale. Otherwise. You may want to enter reasonable starting values for parameter estimates: − For Weibull. For small or heavily censored sample. you can use historical values for distribution parameters to improve your analysis. MLE will tend to overestimate the low percentiles. see Drawing conclusions when you have few or no failures. Maximum likelihood (MLE) • • • Distribution parameter estimates are more precise than least squares (XY). In this case. In Set threshold at. MLE tends to overestimate the shape parameter for a Weibull distribution and underestimate the scale parameter in other distributions. When possible. Least squares estimates are calculated by fitting a regression line to the points in a probability plot. you may want to use the more conservative estimates or consider the advantages of both approaches and make a choice for your problem. When Minitab estimates the parameters using the maximum likelihood method. • Change the maximum number of iterations for reaching convergence (the default is 20). − For other 2-parameter distributions. scale. if the results are consistent. − For exponential. Or. If the maximum number of iterations is reached before convergence. Estimate at least one parameter by: − Fixing the shape OR scale − Fixing the threshold − Fixing the shape OR scale AND threshold Drawing conclusions when you have few or no failures When you have few or no failures. both methods should be tried. . The maximum likelihood estimation method has attractive mathematical qualities. When there is only one failure and some rightcensored observations. Therefore. and threshold. The line is formed by regressing time to failure or log (time to failure) (X) on the transformed percent (Y). The likelihood function describes. For more information.Reliability and Survivial Analysis • • 3 In Set shape (slope−Weibull) or scale (1/slope−other dists) at. enter the threshold value. Estimating the distribution parameters You can choose to estimate the parameters using either the least squares (XY) method or the maximum likelihood method (modified Newton-Raphson algorithm). Minitab obtains maximum likelihood estimates through an iterative process. for each set of distribution parameters. the algorithm terminates. enter the shape or scale value. All rights reserved. you can fix all parameter or you can estimate at least one parameter. Providing historical parameters makes the resulting analysis more precise. MLE allows you to perform an analysis when there are no failures. Least squares estimates versus maximum likelihood estimates Least squares estimates are calculated by fitting a regression line to the points in a probability plot. you can: • Enter starting values for the algorithm. The line is formed by regressing time to failure or log (time to failure) (X) on the transformed percent (Y). Here are the major advantages of each method: Least squares (LSXY) • • Better graphical display to the probability plot because the line is fitted to the points on a probability plot. The maximum likelihood solution may not converge if the starting estimates are not in the neighborhood of the true solution. all results − such as the percentiles − are based on the parameters you enter. then there is more support for your conclusions. enter shape. When you set historical parameters. − For other 3-parameter distributions. enter location. enter mean. enter scale and threshold. scale. no estimation is done. enter location and scale. you can set your own parameters. Click OK. if you like. if your values are an appropriate choice. LSXY is more accurate than MLE. − For 2-parameter exponential. Maximum likelihood estimates are calculated by maximizing the likelihood function. the maximum likelihood parameter estimates may exist for a Weibull distribution. the chance that the true distribution has the parameters based on the sample. − For 3-parameter Weibull. 88 Copyright © 2003–2005 Minitab Inc.

and cumulative failure probabilities. 89 . enter the test value. Note For example. Test for equal scale (Weibull or expo) or location (other distributions): Check to test whether two or more samples have the same scale or location. location. If the lower confidence bound for the 5th percentile is 13. If you collect life data and have no failures. 3 Copyright © 2003–2005 Minitab Inc. Do one or more of the following: • In Test shape (slope−Weibull) or scale (1/slope−other dists) equal to.Distribution Analysis If your data come from a Weibull or exponential distribution. • In Test threshold equal to. For 3-parameter distributions. The maximum likelihood method will be used to estimate parameters. If your data come from a three−parameter Weibull or two−parameter exponential. Click OK. Equality of Parameters Test for equal shape (slope−Weibull) or scale (1/slope−other distributions): Check to test whether two or more samples have the same shape or scale. check all Equality of Parameters options. enter the test value. Minitab automatically assigns a shape parameter of 1. You provide a historical value for the shape parameter (Weibull). or threshold) are consistent with specified values Whether the sample comes from the historical distribution Whether two or more samples come from the same population Whether two or more samples share the same shape. If your data are from an exponential distribution. click Test. and whether two or more samples share the same shape. Parametric Distribution Analysis − Test Stat > Reliability/Survival > Distribution Analysis (Right Censoring) or Distribution Analysis (Arbitrary Censoring) > Parametric Distribution Analysis > Test Tests whether distribution parameters for a sample equal specified values. Note For 2-parameter distributions. scale. location. you must also provide a historical value for the threshold parameter. Minitab can still analyze when all of the following are met: • • • • The data come from a Weibull or exponential distribution. or threshold parameters Dialog box items Consistency of Sample with Value Test shape (slope−Weibull) or scale (1/slope−other dists) equal to: Enter a test value to compare the sample's shape (Weibull) or scale (other distributions). The data are right-censored. See Demonstration Test Plans to determine the optimal testing time or number of test units to use. To compare distribution parameters to a specified value 1 2 In the main dialog box. Minitab performs Wald Tests [7] and provides Bonferroni 95. • In Test scale (Weibull or expo) or location (other dists) equal to. enter the test value. shape. percentiles. location. Test threshold equal to: Enter a test value to compare the sample's threshold. or threshold parameters.0% confidence intervals for the following hypothesis tests: • • • • Whether the distribution parameters (scale. Test scale (Weibull or expo) or location (other dists) equal to: Enter a test value to compare the sample's scale (Weibull or exponential) or location (other distributions). to test the same. and then examine the lower confidence bound to substantiate that the product is at least as good as specifications require. survival probabilities. you conclude that your product meets specifications and terminate the test. you can do a Bayes analysis to obtain lower confidence bounds for parameters. All rights reserved. scale. check the first two Equality of Parameters options to test whether two or more samples come from the same population. your reliability specifications require that the 5th percentile is at least 12 months. You run a Bayes analysis on data with no failures. Test for equal threshold: Check to test whether two or more samples have the same threshold.1 months.

• In Test scale (Weibull or expo) or location (other dists) equal to. • Check Test for equal threshold. All rights reserved. Average (median) of tied points: Choose to plot the median cumulative percent of identical failure times. X axis label: To replace the default x-axis label. • Check Test for equal threshold. enter the parameter of a historical distribution. and cumulative failure plots. Handle tied failure times by plotting All points: Choose to plot the cumulative percents for each occurrence of a failure time. click Graphs. cumulative failure plot. • In Test threshold equal to. click Test. Click OK. Maximum X scale: Enter a value for the maximum x-axis scale. . Click OK. Cumulative failure plot: Check to display a cumulative failure plot. 3 Parametric Distribution Analysis (Right Censoring) − Graphs Stat > Reliability/Survival > Distribution Analysis (Right Censoring) > Parametric Distribution Analysis > Graphs You can draw a probability plot. Do the following: • Check Test for equal shape (slope−Weibull) or scale (1/slope−other dists). Minimum X scale: Enter a value for the minimum x-axis scale. Note To change the method for calculating probability plot points. Survival plot: Check to display a survival plot. Show graphs of different variables or by levels: Choose to display the graphs overlaid on the same graph or on separate graphs. click Test.Reliability and Survivial Analysis To compare parameters from two or more distributions 1 2 In the main dialog box. click Test. Hazard plot: Check to display a hazard plot. To draw a parametric survival plot 1 In the main dialog box. enter the parameter of a historical distribution. Click OK. Do one or more of the following: • Check Test for equal shape (slope−Weibull) or scale (1/slope−other distributions). see Tools > Options > Individual Graphs > Probability Plots. Maximum of tied points: Choose to plot the maximum cumulative percent of identical failure times. and a hazard plot. enter the parameter of a historical distribution. Display confidence intervals on above plots: Check to display confidence intervals on the probability. survival. Do the following: • In Test shape (slope−Weibull) or scale (1/slope−other dists) equal to. 90 Copyright © 2003–2005 Minitab Inc. a survival plot. type the text in this box. • Check Test for equal scale (Weibull or expo) or location (other distributions). 3 To determine whether two or more samples come from the same population 1 2 In the main dialog box. Dialog box items Probability plot: Check to display a probability plot. • Check Test for equal scale (Weibull or expo) or location (other distributions). 3 To test whether a sample comes from a historical distribution 1 2 In the main dialog box.

Specify the method used to obtain the plot points Specify a minimum and/or maximum value for the x-axis scale Enter a label for the x-axis 3 4 5 Click OK. Copyright © 2003–2005 Minitab Inc.Distribution Analysis 2 3 Check Survival plot. (See To change confidence levels if you want to change from 95% confidence intervals. enter a label for the x-axis. click Estimate. enter a value. In Confidence level. Choose Overlaid on the same graph to have multiple samples plotted on the same graph.0% confidence interval for the curve. Click OK. Cumulative failure plots Cumulative failure plots display the cumulative failure probabilities versus time. To change the method used to obtain the fitted line. If you like. In Minimum X scale or Maximum X scale. The cumulative failure curve is surrounded by two outer lines − the approximate 95.) • In Minimum X scale and Maximum X scale. or On separate graphs to have each plot displayed separately. In separate panels on the same graph to have multiple samples plotted in separate panels all on the same graph. which provide reasonable values for the "true" cumulative failure function. See Tools > Options > Individual Graphs > Probability Plots to choose Median Rank (Benard). 4 To modify the default probability plot 1 2 In the main dialog box. click Estimate. To change the confidence level for the 95. Uncheck Display confidence intervals on above plots. Maximum of the tied points..0% confidence interval to some other level. Click OK. Mean Rank (Herd−Johnson). Each plot point represents the cumulative percentage of units failing at time t.0% confidence interval. type values for the x-axis scale.. Click OK.0% confidence interval Choose how plots are displayed For Parametric Distribution Analysis (Right Censoring) Under Handle tied failure times by plotting. In X axis label.. or Average (median) of tied points. Choose what to plot when you have tied failure times Turn off the 95. 91 . do any of the following: • Uncheck Display confidence intervals on above plots to turn off the 95.. All rights reserved. enter values for the scale minimum and maximum. • In X axis label. modified Kaplan−Meier (Hazen). choose All points (default). choose Least Squares (default) or Maximum Likelihood. In Estimation Method. or Kaplan−Meier method. To. Do any of the following: Do. click Graphs. type a label.

The end of the curve. the hazard rate is high at the beginning of the plot. where the failure rate is low. is the normal life stage. is the wearout stage. The early period with high failure rate is often called the infant mortality stage. where failure rate increases again. 92 Copyright © 2003–2005 Minitab Inc. the curve often resembles the shape of a bathtub. The middle section of the curve. . This particular example does not have the bathtub shape. then high again at the end of the plot. Thus.Reliability and Survivial Analysis Hazard plots − parametric distribution analysis The hazard plot displays the instantaneous failure rate for each time t. All rights reserved. low in the middle of the plot. Often.

The Weibull probability plot below shows failure times associated with running engine windings at a temperature of 80° C: Copyright © 2003–2005 Minitab Inc. The associated probabilities are then transformed and used as the y-variables. All rights reserved. To choose from various methods to estimate the plot points. To make the fitted line. The transformed scales. differ depending on the distribution used. see Tools > Options > Individual Graphs > Probability Plots. The percentiles may be transformed. however. So you can use the probability plot to assess whether a particular distribution fits your data. Confidence intervals. differs depending on the parametric distribution chosen. 93 . In general. the closer the points fall to the fitted line. To choose from various methods to obtain the fitted line. • • For more information on probability plot calculations. see Parametric Distribution Analysis − Estimate. The fitted line. see Methods and formulas . depending on the distribution.Distribution Analysis To draw a parametric hazard plot.0% confidence intervals for the fitted line.parametric distribution analysis. Tip To quickly compare the fit of up to eleven different distributions at once. The observed failure times are plotted on the x-axis vs. based on the chosen distribution. Minitab provides two goodness of fit measures to help assess how the distribution fits your data. and are used as the x-variables. set of approximate 95. Because the plot points do not depend on any distribution. which is a graphical representation of the percentiles. chosen to linearize the fitted line. the better the fit. which represent the proportion of failures up to a certain time. they would be the same (before being transformed) for any probability plot made. check Hazard plot in the Graphs subdialog box. Minitab first calculates the percentiles for the various percents. Transformations of both the x and y data are needed to ensure that the plotted y values are a linear function of the plotted x values if the data are sampled from the particular distribution. the estimated cumulative probabilities (p) on the y-axis. Minitab calculates the plot points using a nonparametric method. The plot consists of: • Plot points. Fitted line. Probability plots Use a probability plot to assess whether a particular distribution fits your data. see Distribution ID Plot (Right Censoring) or Distribution ID Plot (Arbitrary Censoring).

0% confidence interval for the curve. 94 Copyright © 2003–2005 Minitab Inc.Reliability and Survivial Analysis Survival plots Survival (or reliability) plots display the survival probabilities versus time. which provide reasonable values for the "true" survival function. All rights reserved. Each plot point represents the proportion of units surviving at time t. The survival curve is surrounded by two outer lines − the approximate 95. .

and failure mode represents a unique distribution that requires estimates in one single column or in separate columns. Use historical estimates: Enter one column of values to be used for all samples. By variable group level. Minitab also displays them. Click OK. Dialog box items Control the Display of Results Display nothing: Choose to suppress all Session window output. See Starting estimates for more information. and tests of parameters. enter columns of starting values based on your data structure. If Least Squares is your estimation method. and percentiles. − − 4 5 In Use starting estimates. table of percentiles and survival probabilities: Choose to also display the MTTF. Parametric Distribution Analysis − Options Stat > Reliability/Survival > Distribution Analysis (Right Censoring) or Distribution Analysis (Arbitrary Censoring) > Parametric Distribution Analysis > Options You can specify whether the distribution parameters should be calculated from the data or enter your own historical estimates. In Maximum number of iterations. Use historical estimates. median. to enter your own estimates for the distribution parameters. Show log-likelihood for each iteration of algorithm: Check to show the log-likelihood for each iteration of algorithm. See the following table for various scenarios: Copyright © 2003–2005 Minitab Inc. you can do any of the following: Starting estimates and historical estimates You must enter starting estimates or historical estimates for each distribution of failures each parameter you wish to estimate. parameter estimates. If you choose Use historical estimates. Variable information. In addition. choose one of the following: • • 3 Estimate parameters of distribution. The default number of iterations is 20. to estimate the distribution parameters from the data. choose Use historical estimates. standard deviation. enter a positive integer to specify the maximum number of iterations for estimation. Use starting estimates: Enter starting estimates for the parameters. See Historical estimates for more information. Each variable. estimated parameters. If Maximum Likelihood is your estimation method. If you select survival probabilities in the Estimate subdialog box. Minitab also displays the assumed distribution. censoring information. All rights reserved. log-likelihood. If you choose Estimate parameters of distribution. Display analyses for individual failure modes according to display of results: Check to display the results chosen above for each failure mode. or several columns of values that match the order in which the corresponding variables appear in the Variables box in the main dialog box. To control estimation of the parameters 1 2 In the main dialog box. Maximum number of iterations: Enter the maximum number of iterations. or several columns of values which match the order in which the corresponding variables appear in the Variables box in the main dialog box. Enter one column of values to be used for all samples. goodness-of-fit and tests of parameters: Choose to display censoring information and the estimation method. 95 . quartile information. − Enter columns of historical values based on your data structure. characteristics of distribution.Distribution Analysis Parametric Distribution Analysis − Results Stat > Reliability/Survival > Distribution Analysis (Right Censoring) or Distribution Analysis (Arbitrary Censoring) > Parametric Distribution Analysis > Results You can control the display of Session window output. click Options. Dialog box items Estimate parameters of distribution: Choose to estimate the distribution parameters from the data (the default).

Enter a single column containing values for each parameter for each failure mode of each group level. If there are four group levels. variable 1. Enter one column of values for each group level. Available when you estimate survival probabilities or cumulative failure probabilities at different times. If there are two variables with two failure modes. Enter • • • A By variable with several group levels. 96 Copyright © 2003–2005 Minitab Inc. If there are two group levels with three failure modes each. enter six columns of parameter estimates. you can store information pertaining to the overall distribution only. • • Variables with multiple failure modes: Case A single variable with a multiple failure modes. Available when you estimate survival probabilities at different times. Two or more variables. If variable 2 does not have failure mode 1. the second column contains the parameter estimates of failure mode 2. Enter a single column containing values for each parameter for each variable. etc. variable 1. enter the number of levels the column of grouping indicators contains. the second column − parameter estimates of failure mode 1. Enter one column of parameter estimates for each failure mode of each group level. Enter one column of parameter estimates for each variable. Enter a single column containing values for each parameter for each failure mode and variable. Characteristics of Fitted Distribution Percentiles: Check to store percentiles. Available when you estimate survival probabilities at different times. Dialog box items Enter number of levels in by variable If all of the samples are stacked in one column. enter four columns containing parameter estimates. Times for probabilities: Check to store times for survival probabilities and cumulative failure probabilities. • • A By variable with several group levels and multiple failure modes. Percents for percentiles: Check to store percents for percentiles. • • Parameter estimates are assigned by failure mode then variable. Parametric Distribution Analysis − Storage Stat > Reliability/Survival > Distribution Analysis (Right Censoring) or Distribution Analysis (Arbitrary Censoring) > Parametric Distribution Analysis > Storage You can store characteristics of the fitted distribution and information on the distribution parameters in your worksheet. Enter one column containing values for each parameter of your distribution.Reliability and Survivial Analysis Variables with single failure modes: Case A single variable with a single failure mode. All rights reserved. Normally. the first column should contain the parameter estimates of failure mode 1. Confidence limits for percentiles: Check to store confidence limits for percentiles. Enter a single column containing values for each parameter for all group levels. Survival probabilities: Check to store survival probabilities. If there are three failure modes. Confidence limits for survival probabilities: Check to store confidence limits for survival probabilities. Two or more variables. Standard error of percentiles: Check to store standard error of percentiles. Enter one column of parameter estimates for each failure mode of each variable. Enter • • Enter a single column containing values for each parameter for each failure mode. enter four columns of parameter estimates. . each with a single failure mode. each with multiple failure modes. enter three columns of parameter estimates. Caution Do not enter starting estimates or historical estimates for failure modes that you have eliminated from the analysis. Enter one column of parameter estimates for each failure mode. With multiple failure modes. variable 2.

80° C and 100° C.MTW. From Assumed distribution. Variance/covariance matrix: Check to store the variance / covariance matrix. Click Censor. You decide to look at failure times for engine windings at two temperatures. You also want to draw two plots: a probability plot to see if the lognormal distribution provides a good fit for your data. Click OK in each dialog box. Session window output Distribution Analysis: Temp80 Variable: Temp80 Censoring Information Uncensored value Right censored value Count 37 13 Censoring value: Cens80 = 0 Estimation Method: Least Squares (failure time(X) on rank(Y)) Distribution: Lognormal Parameter Estimates Standard Error 0.503340 Parameter Location Scale Estimate 4. You want to find out the following information for each temperature: • • Times at which various percentages of the windings fail. In Variables. you collect failure times (in months) for 50 windings at 80° C. you collect failure times for 40 windings at 100° C. and a survival plot. Click OK.339626 0. enter 0. Choose Stat > Reliability/Survival > Distribution Analysis (Right Censoring) > Parametric Distribution Analysis. Example of a parametric distribution analysis with exact failure/right censored data Suppose you work for a company that manufactures engine windings for turbine assemblies. Log-likelihood for last iteration: Check to store the log-likelihood ratio for the last iteration.03430 0.91671 4.15189 0. Available when you estimate cumulative failure probabilities at different times. All rights reserved.827 Copyright © 2003–2005 Minitab Inc. Confidence limits for parameters: Check to store confidence limits for parameters. 97 . enter 70. Choose Use censoring columns and enter Cens80 Cens100 in the box. choose Lognormal.0414962 95.0% Normal CI Lower Upper 3. In the first sample. Confidence limits for cumulative failure probabilities: Check to store confidence limits for cumulative failure probabilities.413458 Log-Likelihood = -182. Click Estimate. In Estimate probabilities for these times (values). Check Survival plot.0599960 0. you use a column of censoring indicators to designate which times are actual failures (1) and which are censored units removed from the test before failure (0). Available when you estimate cumulative failure probabilities at different times. Some of the windings drop out of the test for unrelated reasons. You are particularly interested in the 0. in the second sample. Click OK. Information on Parameters Choose to store the following when you have a single failure mode. In Estimate percentiles for these additional percents. Standard error of estimates: Check to store standard error of parameter estimates. Proportion of windings that survive past 70 months. 1 2 3 4 5 6 7 8 Open the worksheet RELIABLE. In the Minitab worksheet. Engine windings may decompose at an unacceptable rate at high temperatures.1. Parameters estimates: Check to store parameter estimates.1st percentile. enter Temp80 Temp100.Distribution Analysis Cumulative failure probabilities: Check to store cumulative failure probabilities. Click Graphs.

78623 95.53192 2.6233 29.68093 2.65838 2.206353 0.6064 32.6850 21.Reliability and Survivial Analysis Goodness-of-Fit Anderson-Darling (adjusted) = 67.009 107.0819 10.9780 25.4895 50.2348 55.5452 26.968 132.535 Percent 0.2253 34.9866 34.3031 40.8979 45.1712 25.792 118.10502 3.46626 2.0% Normal CI Lower Upper 0.982 Characteristics of Distribution Standard Error 3.8841 56.539 116.158 140.40509 8.9584 69.1 1 2 3 4 5 6 7 8 9 10 20 30 40 50 60 70 80 90 91 92 93 94 95 96 97 98 99 Percentile 15.3858 22.2788 Mean(MTTF) Standard Deviation Median First Quartile(Q1) Third Quartile(Q3) Interquartile Range(IQR) Estimate 61.86871 9.2361 35.5033 42.1482 50.4791 69.597 133.851 184.5197 32.3610 101.38997 2.41666 10.3418 36.5031 93.9629 27.928 127.493 107.52897 7.2638 19. .371 118.5298 48.65393 8.0947 83.446 20.461 111.6834 79.025 148.2708 38.9145 50.9246 Table of Percentiles Standard Error 2.414111 Time 70 Probability 0.6247 84.23848 2.70119 3.7095 30.4305 96.8060 27.5736 56.8591 85.0014 90.6957 31.7560 23.63147 2.0421 13.61498 2.1368 38.988 161.00485 8.5730 57.9307 24.084 147.92345 3.7018 65.81469 4.3487 63.67001 2.7224 95.6244 112.84200 4.6798 31.905 122.5539 37.5279 20.8003 40.9548 45.7422 51.8029 27.9744 101.4020 34.7524 74.842 Table of Survival Probabilities 95.012 104.0% Normal CI Lower Upper 54.6947 61.5033 62.9209 12.78266 2.528 122.1837 80.3978 28.220 115.3448 37.56907 2.8113 88. All rights reserved.4582 33.5948 24.59497 2.5010 91.7465 21.0% Normal CI Lower Upper 11.6251 28.38997 3.9173 17.6770 31.46368 5.302208 Distribution Analysis: Temp100 Variable: Temp100 Censoring Information Uncensored value Right censored value Count 34 6 Censoring value: Cens100 = 0 Estimation Method: Least Squares (failure time(X) on rank(Y)) 98 Copyright © 2003–2005 Minitab Inc.8849 82.0123 29.9747 25.1236 45.9245 26.0198 95.9826 98.91315 3.5539 70.82954 3.7430 70.656 Correlation Coefficient = 0.2348 63.6963 16.2625 39.64569 2.

2774 52.15631 9.9469 18.0158 69.675 107.1848 13.8260 56.0826 92.1 1 2 3 4 5 6 7 8 9 10 20 30 40 50 60 70 80 90 91 92 93 94 95 96 97 98 99 Percentile 3.5006 21.0101 25.78631 6.28227 12.27898 2.2830 16.61936 0.1262 23.0471 7.3978 24.029 70.93927 19. 99 .114 136.7690 32.0951 25.6006 71.06524 6.8253 80.7263 26.601 90.4474 20.3710 100.23760 16.438 80.5444 47.0% Normal CI Lower Upper 2.3841 173.9360 6.26891 10.5534 28.4820 38.38589 3.016 346.740387 Log-Likelihood = -160.9574 68.119118 0.1294 29.02085 2.4885 29.54010 7.39602 18.8656 24.653 76.4651 19.0949534 95.0123 55.8015 12.2080 11.188 97.0798 41.276 117.658 126.2247 206.0401 11.60714 16.394 150.3603 231.5013 9.0% Normal CI Lower Upper 37.2146 40.689 105.17097 1.85283 0.0% Normal CI Lower Upper 3.32954 2.8278 14.16701 13.5444 47.0222 8.33100 95.9453 15.1275 19.294 73.5027 29.2232 Mean(MTTF) Standard Deviation Median First Quartile(Q1) Third Quartile(Q3) Interquartile Range(IQR) Estimate 49.2924 42.3137 45.2750 136.3075 30.697 Goodness-of-Fit Anderson-Darling (adjusted) = 17.3643 57.1261 35.2393 26.92520 15.16476 2.6459 61.2744 19.7518 143.704 208. All rights reserved.80788 1.1261 17.92809 2.8572 95.7800 187.27058 10.22448 2.98322 11.9364 37.575829 0.197736 Copyright © 2003–2005 Minitab Inc.94163 4.153 84.5941 6.880 Table of Survival Probabilities 95.5806 96.Distribution Analysis Distribution: Lognormal Parameter Estimates Standard Error 0.323727 Time 70 Probability 0.4516 39.44473 2.8361 Table of Percentiles Standard Error 1.13270 10.54241 7.600 111.44473 5.7576 33.1357 64.978 126.78063 3.83086 17.2377 10.4642 71.281 Correlation Coefficient = 0.233 Percent 0.951972 Parameter Location Scale Estimate 3.9073 8.09791 2.5320 151.62573 1.7510 5.988 Characteristics of Distribution Standard Error 6.66562 8.0% Normal CI Lower Upper 0.186 170.107086 0.663 270.5122 13.4125 4.9318 37.1446 53.71014 4.3137 22.18917 3.7007 161.37705 2.97661 8.7282 9.3003 46.

look at the Table of Percentiles. 100 Copyright © 2003–2005 Minitab Inc. .Reliability and Survivial Analysis Graph window output Interpreting the results To see the times at which various percentages of the windings fail. it takes 21. At 80° C.5948 months for 1% of the windings to fail. All rights reserved. for example.

5866 68.Log-likelihood and goodness-of-fit statistics Characteristics of distribution and their .22% survive past 70 months.3032 56.872053 2. 0.0996 51.7777 44.7720 39.4674 68.0% Normal CI Lower Upper 51. At 80° C.78798 1.26856 57.78631 months.0% Normal CI Lower Upper 3.Standard errors .1st percentile. 0.95% confidence intervals Probability plot for each failure mode • • • For example.4415 74.95% confidence intervals Table of percentiles and their .73936 0.Distribution Analysis You can find the 0. All rights reserved.27922 95.606 Correlation Coefficient = 0.5 months.Standard errors . What proportion of windings would you expect to still be running past 70 months? In the Table of Survival Probabilities you find your answer. 30.6757 19.8914 63.7263 26. At 80° C. So the increase in temperature decreased the percentile by about 9.10091 4. 19.7498 Mean(MTTF) Standard Deviation Median First Quartile(Q1) Third Quartile(Q3) Interquartile Range(IQR) Estimate 56.Standard errors . Distribution Analysis: Temp80 Variable: Temp80 Censoring Information Uncensored value Right censored value Count 37 13 Censoring value: Cens80 = 0 Estimation Method: Least Squares (failure time(X) on rank(Y)) Distribution: Weibull Parameter Estimates Standard Error 0. within the Table of Percentiles.1075 Copyright © 2003–2005 Minitab Inc.00877 2.986 Characteristics of Distribution Standard Error 2.6177 Parameter Shape Scale Estimate 3.84080 95.6179 17.0629 50.258 Goodness-of-Fit Anderson-Darling (adjusted) = 67.1% of the windings fail by 3. which you requested.7956 Log-Likelihood = -202. at 100° C.2497 15.3558 62.63819 62.4955 62. at 100° C.296611 2.90747 3.77% survive. 101 .95% confidence intervals .6941 24.7465 months.1% of the windings fail by 15. Output − Parametric Distribution Analysis The default output for Parametric Distribution Analysis (Right Censoring) and Parametric Distribution Analysis (Arbitrary Censoring) is: • • Censoring information Parameter estimates and their .3817 21.

01533 3.8988 86.0826 71.2089 56.01453 2.5509 46.97051 3. you can request additional output and display Session window output for each failure mode. To make a quick Kaplan-Meier survival plot and empirical hazard plot. When you have multiple samples.0717 47.5364 101.2342 30.7796 53.4397 83. and draw survival.7063 86. When you have tabled data with a varied censoring scheme.4656 89.1542 28.8624 78.3374 19.94029 2.6859 33.1636 83.1039 32.1676 60.6796 81. When you have exact failure/right-censored data.99523 2.8299 41. and hazard plots.Right Censoring also tests the equality of their survival curves.5499 From the Results subdialog box.8152 33.0814 89.9534 67. When you have exact failure/right-censored data and multiple samples.5920 88.44778 2.0525 26. which perform the full analysis. 102 Copyright © 2003–2005 Minitab Inc. Nonparametric Distribution Analysis .5795 47.4812 40.4857 24.0423 27. cumulative failure probabilities. Minitab tests the equality of survival curves.4813 25.7228 32. hazard rates.2010 22.7720 55.0778 21.4461 84.9618 24.85684 2.74337 2.8274 26.8233 66.2431 16.9520 81.78019 2. Gives you nonparametric estimates of the survival probabilities.6000 85.89222 2.2256 30.85336 2.967 Percent 1 2 3 4 5 6 7 8 9 10 20 30 40 50 60 70 80 90 91 92 93 94 95 96 97 98 99 Note Percentile 17. Nonparametric Distribution Analysis Nonparametric distribution analysis commands The nonparametric distribution analysis commands include Nonparametric Distribution Analysis-Right Censoring and Nonparametric Distribution Analysis-Arbitrary Censoring.77225 2.3384 77.2159 73.78771 2. and other functions.95551 2.79777 2.8345 87.95707 2.6536 91.0809 94. This graph is often used before the full analysis to view summary information.3174 36.7339 21.5024 39. you can request Turnbull or Actuarial estimates.81118 2.9283 41.7259 58. hazard rates.6375 97.85714 2.0084 82.Reliability and Survivial Analysis Table of Percentiles Standard Error 2.4531 38.90747 2.0% Normal CI Lower Upper 13.6162 74. cumulative failure.1142 92.8008 71. see Distribution Overview Plot (Right Censoring). or an Actuarial survival plot and hazard plot.3351 51. Command Distribution Overview Plot Right Censored Arbitrary Censored Nonparametric Distribution Analysis Right Censored Arbitrary Censored Description Draws a Kaplan-Meier survival plot and hazard plot on one graph. and create a Distribution Overview Plot.6817 76.4865 90.5305 23. on one graph. .5348 31.7571 29. cumulative failure.0685 27. you can request Kaplan-Meier or Actuarial estimates.6707 85.1830 36.6068 75. and other estimates depending on the nonparametric technique chosen.8866 27.93652 3. and hazard plots.3005 52.5708 78. All rights reserved. You can estimate survival probabilities.82909 2. and draw survival. cumulative failure probabilities. Draws a Turnbull survival plot.16895 95. Nonparametric Distribution Analysis (Right Censoring) Stat > Reliability/Survival > Distribution Analysis (Right Censoring) > Nonparametric Distribution Analysis Use this command when you have right censored data or actual failure times and no distribution fits your data.3558 62.81001 2.3046 66.0728 35.1794 79.7094 84.80866 2.3603 95.88716 2.7777 61.91916 2.64045 2.9748 79.

the second censoring column is paired with the second data column. Singly censored data are more common in controlled studies. Note Occasionally. enter the columns in Frequency columns. so you can compare the various functions between samples. and enter a column of grouping indicators in the box Distribution Analysis (Right Censored Data) If your data include exact failures or if test units do not fail before your study is over. For systems that have more than one cause of failure. see Distribution Analysis Data. Frequency columns (optional): Enter the columns of frequency data. you must have a column of censoring indicators. 7 If you like. Multiply censored data are more common in the field. Minitab can analyze systems with one cause of failure or multiple causes of failure. where units go into service at different times. For example. use Parametric Distribution Analysis (Right Censoring). If you like. • For data with censoring columns: Choose Use censoring columns. • • For time censored data: Choose Time censor at. See Drawing conclusions when you have few or no failures. If you have no censored values. entering 500 says that any observation from 500 time units onward is considered censored. enter the value you use to indicate a censored value in Cens value. You can enter up to 50 samples per analysis. you may have life data with no failures. In Variables. You can enter up to 50 columns (50 different samples). with different colors and symbols. All rights reserved. then enter a number of failures at which to begin censoring. Dialog box items Variables: Enter the columns of failure times. your data are right-censored. For general information on life data and censoring. See Worksheet Structure for more information. The first censoring column is paired with the first data column. then enter a failure time at which to begin censoring. Censor Stat > Reliability/Survival > Distribution Analysis (Right Censoring) > distribution analysis command > Censor Allows you to designate which observations are censored. For example. • When your data are multiply censored. Right censored data can be: • Singly censored − All of the test units run for the same amount of time. Failed units are considered exact failures. then enter the censoring columns in the box. then click OK. Minitab uses the lowest value in the censoring column by default. see Multiple Failure Modes (Right Censored Data). check By variable. enter a column of grouping indicators in By variable. you can skip steps 5 & 6. Under certain conditions. entering 150 says to censor all (ordered) observations starting with the 150th observation censored. If all of the samples are stacked in one column. All of the samples display on a single plot. and so on.Distribution Analysis If a distribution fits your data. Note 5 6 Click Censor. If you have frequency columns. Minitab estimates the functions independently for each sample. 103 . Do one of the following. For failure censored data: Choose Failure censor at. then click OK. Units surviving at the end of the study are considered censored data. Copyright © 2003–2005 Minitab Inc. You can enter up to 50 columns (50 different samples). By variable: If all of the samples are stacked in one column. Failure times are intermixed with censoring times. and leave all other observations uncensored. If you do not enter a censoring value. enter the columns of failure times. use any of the available dialog box options. To do a nonparametric distribution analysis (uncensored/right censored data) 1 2 3 4 Choose Stat > Reliability/Survival > Distribution Analysis (Right Censoring) > Nonparametric Distribution Analysis. Multiply censored − Test units are censored at different times. Minitab allows you to draw conclusions based on that data. unless you assume a common shape (Weibull) or scale (other distributions).

Reliability and Survivial Analysis Dialog box items Censoring Options Use censoring columns: Choose Use censoring columns. then enter a failure time at which to begin censoring. Nonparametric Distribution Analysis (Right Censoring) − Estimate Stat > Reliability/Survival > Distribution Analysis (Right Censoring) > Nonparametric Distribution Analysis > Estimate You can choose the method of estimation for the nonparametric estimates of survival probabilities. 10-20. and so on. For example. If you do not enter a value. Confidence level: Enter a confidence level for all of the confidence intervals. Dialog box items Estimation Method Kaplan-Meier: Choose to estimate the parameters using the Kaplan-Meier method. Time censor at: For time censored data. Dialog box items Use failure mode columns: Enter the columns containing the failure modes. Failure Mode Options Use all failure modes: Choose to include all failure modes in the analysis. entering 150 says to make all (ordered) observations starting with the 150th observation censored. Text values must be contained in double quotes. entering 500 says that any observation from 500 time units onward is considered censored. 104 Copyright © 2003–2005 Minitab Inc. Minitab uses the lowest value in the censoring column. The default is 95. Midpoint of intervals: Choose if a failure for any one failure mode causes the experiment to end. All rights reserved. Estimate survival probabilities: Choose to estimate survival probabilities. the second censoring column is paired with the second data column. The first censoring column is paired with the first data column. in the box. Failure censor at: For failure censored data. Estimate cumulative failure probabilities: Choose to estimate cumulative failure probabilities. Create right censored observations using (Only available with arbitrary censoring. Actuarial method: Choose to estimate the parameters using the actuarial method. choose Failure censor at. For example. For example. Specify time intervals as 0 to __ by __: Choose to use equally spaced time intervals − choose and enter numbers in the boxes. etc. 68. and so on up to 80-100. gives you these time intervals: 0-4. or a column of numbers. For example. 0 to 100 by 20 gives you these time intervals: 0-20. Confidence intervals: Choose to use two-sided confidence intervals (the default) or just an upper or lower confidence interval. then enter a number of failures at which to begin censoring. Enter endpoints of intervals: Choose to use unequally spaced time intervals and enter a series of numbers. 20-40.0%. and all other observations uncensored. Eliminate failure modes: Enter the failure modes to exclude from the analysis. Nonparametric Distribution Analysis − Failure Mode Stat > Reliability/Survival > Distribution Analysis (Right Censoring) or Distribution Analysis (Arbitrary Censoring) > Nonparametric Distribution Analysis > FMode Use to estimate the overall reliability of your system when multiple causes of failure exist in order to investigate the reliability of the individual failure modes. entering 0 4 6 8 10 20 30. hazard rates. . 8-10. 4-6. choose Time censor at.) Use to determine how Minitab will create right-censored observations for other failure modes when data are interval censored. enter the value you use to indicate censoring in Censoring value. This option is available for both right-censored and arbitrarily censored data. Right endpoint of intervals: Choose if the experiment continues until the right endpoint when a failure in the interval occurs. and 20-30. Use failure modes: Enter the failure modes to include in the analysis. then enter the censoring columns in the box. cumulative failure probabilities. Censoring value: If you like.

entering 0 4 6 8 10 20 30. 68. where failure rate increases again. click Estimate. click Estimate. 105 . the hazard rate is high at the beginning of the plot. hazard and density estimates (actuarial method). More To change confidence levels 1 2 3 In the main dialog box. choose In addition. More Hazard plots − nonparametric distribution analysis The hazard plot displays the instantaneous failure rate for each time t. in the box. Check Hazard plot. where the failure rate is low. hazard. Often. If you want to plot Actuarial estimates. 4 Click OK. is the wearout stage. then click OK. Dialog box items Survival plot: Check to display a survival plot. and enter a series of numbers. check Actuarial. and 20-30.Distribution Analysis To request actuarial estimates 1 2 3 In the main dialog box. 0 to 100 by 20 gives you these time intervals: 0-20. Do one of the following and then click : With Nonparametric Distribution Analysis (Right Censoring). click Results. gives you these time intervals: 0-4. Display confidence intervals on plot: Check to display confidence intervals on the survival and cumulative failure plots. the hazard plot for Nonparametric Distribution Analysis (Right Censoring) uses the empirical hazard function. Thus. Show graphs of different variables or by levels: Choose to display the graphs overlaid on the same graph or on separate graphs. 8-10. cumulative failure plots. To draw a hazard plot 1 2 3 In the Nonparametric Distribution Analysis (Right Censoring) dialog box. The end of the curve. In Confidence level. enter the desired text in this box. Maximum X scale: Enter a value for the maximum X-axis scale. For example. For example. then high again at the end of the plot. is the normal life stage. choose Actuarial method in the Estimate subdialog box. click Graphs. use any of the available dialog box options. 4-6. The middle section of the curve. or a column of numbers. Cumulative failure plot: Check to display a cumulative failure plot. Nonparametric Distribution Analysis − Graphs Stat > Reliability/Survival > Distribution Analysis (Right Censoring) or Distribution Analysis (Arbitrary Censoring) > Nonparametric Distribution Analysis > Graphs You can draw survival plots. By default. enter a value. 20-40. density (actuarial method) estimates and log-rank and Wilcoxon statistics. Copyright © 2003–2005 Minitab Inc. from the main dialog box. If you like. and hazard plots. 10-20. Click OK. do one of the following: • • use equally spaced time intervals − choose 0 to_by_ and enter numbers in the boxes. the curve often resembles the shape of a bathtub. use unequally spaced time intervals − choose Enter endpoints of intervals. Minimum X scale: Enter a value for the minimum X-axis scale. With Nonparametric Distribution Analysis (Right Censoring). Under Estimation Method. To display hazard and density estimates in the Actuarial table. The early period with high failure rate is often called the infant mortality stage. X axis label: To replace the default X-axis label with your own label. See To request actuarial estimates. Hazard Plots: Check to display a hazard plot. choose In addition. All rights reserved. With Nonparametric Distribution Analysis (Arbitrary Censoring). and so on up to 80-100. low in the middle of the plot.

See Comparing survival curves when you perform nonparametric distribution analysis on more than one right−censored sample. The survival curve is surrounded by two outer lines − the 95% confidence interval for the curve. which provide reasonable values for the "true" survival function. You can interpret the nonparametric survival curve in a similar manner as you would the parametric survival curve. but you can choose to plot Actuarial estimates.Reliability and Survivial Analysis Nonparametric hazard estimates are calculated various ways: • • Nonparametric Distribution Analysis (Right Censoring) automatically plots the empirical hazard function. Nonparametric Distribution Analysis (Arbitrary Censoring) only plots Actuarial estimates. . be sure to choose Actuarial method in the Estimate subdialog box when you want to draw a hazard plot. Each plot point represents the proportion of units surviving at time t. but you can choose to plot Actuarial estimates. You can optionally plot Actuarial estimates. 106 Copyright © 2003–2005 Minitab Inc. the survival plot uses Turnbull survival estimates by default. the survival plot uses Kaplan−Meier survival estimates by default. Since the Actuarial method is not the default estimation method. All rights reserved. depending on the command you use: • • With Nonparametric Distribution Analysis (Right Censoring). You can choose from various estimation methods. With Nonparametric Distribution Analysis (Arbitrary Censoring). The major difference is that the nonparametric survival curve is a step function while the parametric survival curve is a smoothed function. Nonparametric survival plots Survival (or reliability) plots display the survival probabilities versus time.

quartile information. and Wilcoxon statistics. Copyright © 2003–2005 Minitab Inc.7152 Wilcoxon 13. density estimates. standard deviation. In addition. and displays this table in the Session window: Comparison of Survival Curves Test Statistics Method Chi-Square Log-Rank 7. 107 . choose In addition.Distribution Analysis Comparing survival curves (nonparametric distribution analysis-right censoring only) When you enter more than one sample. censoring information. Dialog box items Control the display of results Display nothing: Choose to suppress all Session window output. All rights reserved. characteristics of variable and test statistics for comparing survival curves: Choose to display censoring information and MTTF.0055 0. Kaplan-Meier survival probabilities or actuarial tables: Choose to also display Kaplan-Meier survival probabilities or actuarial tables. and test statistics for survival curves. Variable information. A p-value < α indicates that the survival curves are significantly different. median. Nonparametric Distribution Analysis (Right Censoring) − Results Stat > Reliability/Survival > Distribution Analysis (Right Censoring) > Nonparametric Distribution Analysis > Results You can control the display of Session window output. density (actuarial method) estimates and log-rank and Wilcoxon statistics in the Results subdialog box. hazard. Nonparametric Distribution Analysis-Right Censoring automatically compares their survival curves.1326 DF 1 1 P-Value 0. To get more detailed log-rank and Wilcoxon statistics. hazard.0003 This table contains measures that tell you if the survival curves for various samples are significantly different. density (actuarial methods) estimates and log-rank and Wilcoxon statistics: Choose to also display hazard estimates. log-rank statistics. In addition.

Some of the windings drop out of the test for unrelated reasons. Standard error for survival probabilities: Check to store standard errors for survival probabilities. Click Censor. Confidence limits for cumulative failure probabilities: Check to store confidence limits for cumulative failure probabilities. Check Survival plot and Display confidence intervals on plot. In the Minitab worksheet. enter the number of levels the column of grouping indicators contains. 80° C and 100° C. Survival probabilities: Check to store survival probabilities. in the second sample. 1 2 3 4 5 Open the worksheet RELIABLE. Hazard rates: Check to store the hazard estimates. Cumulative failure probabilities: Check to store cumulative failure probabilities. Dialog box items Enter number of levels in by variable If all of the samples are stacked in one column. you collect times to failure for 50 windings at 80° C. Choose Use censoring columns and enter Cens80 Cens100 in the box. Click Graphs. you collect times to failure for 40 windings at 100° C. Click OK in each dialog box. You decide to look at failure times for engine windings at two temperatures. you use a column of censoring indicators to designate which times are actual failures (1) and which are censored units removed from the test before failure (0). enter Temp80 Temp100. In the first sample. Nonparametric estimates Check any of the items below to store them in the worksheet. Nonparametric Distribution Analysis (Right Censoring) − Storage Stat > Reliability/Survival > Distribution Analysis (Right Censoring) > Nonparametric Distribution Analysis > Storage You can store various probability and rate estimates. All rights reserved. You want to find out the following information for each temperature: • • the times at which half of the windings fail the proportion of windings that survive past various times You also want to know whether or not the survival curves at the two temperatures are significantly different. Standard error for cumulative failure probabilities: Check to store standard errors for cumulative failure probabilities. In Variables. Session window output Distribution Analysis: Temp80 Variable: Temp80 Censoring Information Uncensored value Right censored value Count 37 13 Censoring value: Cens80 = 0 108 Copyright © 2003–2005 Minitab Inc. Times for probabilities: Check to store times for survival probabilities and cumulative failure probabilities. Click OK. Engine windings may decompose at an unacceptable rate at high temperatures. . The output for the 100° C sample follows that of the 80° C sample. Choose Stat > Reliability/Survival > Distribution Analysis (Right Censoring) > Nonparametric Distribution Analysis.MTW. Times for hazard rates: Check to store the times for the hazard estimates. Confidence limits for survival probabilities: Check to store confidence limits for survival probabilities. Example of nonparametric distribution analysis with exact failure/right censored data Suppose you work for a company that manufactures engine windings for turbine assemblies.Reliability and Survivial Analysis Display analyses for individual failure modes according to display of results: Check to display the results for each failure mode. The comparison of survival curves appears last.

0% Normal CI Lower Upper 34.208697 0.97007 0.560000 0.0518459 0.00000 0.0686440 0.641621 0.0701997 0.49305 0.95618 0.99520 0.844803 0.800000 0.420000 0.38029 Distribution Analysis: Temp100 Variable: Temp100 Censoring Information Uncensored value Right censored value Count 34 6 Censoring value: Cens100 = 0 Nonparametric Estimates Characteristics of Variable Standard Error 3.0659697 0.840000 0.6563 Median = 38 IQR = 30 Q1 = 24 Q3 = 54 Copyright © 2003–2005 Minitab Inc.264210 0.0565685 0.689128 0.0543323 0.47130 0.738384 0.360000 0.528697 0.763822 0.340000 0.0277128 0.860000 0.0424264 0.0459565 0.55680 0.500000 0.0707107 0.94162 0.0704840 0.280000 0.0678823 0.980000 0.46953 95.320000 0.0% Normal CI Lower Upper 51.40445 0.3746 60.0254 Mean(MTTF) 55.0706541 0.400000 0.00000 0.0490714 0.0692820 0.0706541 0.0701997 0.780000 0.660000 0.0621592 95.0697997 0.540000 0.0648074 0. 109 .960000 0.0585833 0.65848 0.92649 0.8561 48.440000 0.89482 0.226953 0.136632 0.258462 Standard Error 0.245460 0.905684 1. All rights reserved.941195 1.572980 0.61848 0.700000 0.0383667 0.53579 0.422411 0.4564 Mean(MTTF) 41.283195 0.520000 0.0669925 0.361410 0.0603987 0.580000 0.820000 0.0197990 0.51454 0.760000 0.0634980 0.920000 0.880000 0.155546 0.98315 0.443195 0.480000 0.80930 0.789927 0.401854 0.341521 0.190702 0.87838 0.0659697 0.44930 0.79130 0.816846 0.0669925 0.91087 0.680000 0.71680 0.550702 0.20686 95.380000 0.900000 0.57759 0.67815 0.69759 0.713511 0.Distribution Analysis Nonparametric Estimates Characteristics of Variable Standard Error 2.0697997 0.7 Median = 55 IQR = * Q1 = 48 Q3 = * Kaplan-Meier Estimates Number at Risk 50 49 48 46 45 44 43 42 41 40 39 38 35 34 33 29 28 27 26 25 24 22 21 20 19 18 17 16 13 Time 23 24 27 31 34 35 37 40 41 45 46 48 49 50 51 52 53 54 55 56 58 59 60 61 62 64 66 67 74 Number Failed 1 1 2 1 1 1 1 1 1 1 1 3 1 1 4 1 1 1 1 1 2 1 1 1 1 1 1 2 1 Survival Probability 0.63859 0.665179 0.302411 0.82702 0.381521 0.0% Normal CI Lower Upper 0.

57820 0.38419 0.0774597 0.42500 0.47500 0.926617 1.45000 0.77500 0.95000 0.0765466 0.0344601 0.70000 0.62975 0.72500 0.0684653 0.0632456 0.99297 0.32500 0.0724569 0.92396 0.005 0.0706001 0.34940 0.0522913 0.79781 0.474972 0.179852 0.1326 DF 1 1 P-Value 0.772511 0.0754155 0.30000 0.0706001 0.320245 0.395828 0.55000 0.52500 0.0640434 0.87500 0.0765466 0.0246855 0.67975 0.21875 0.80000 0.843376 1.47015 0.807031 0.0684653 0.0789581 0.23457 Distribution Analysis: Temp80.27500 0.12500 Standard Error 0.0666585 0.60000 0.31302 0.90441 0.Reliability and Survivial Analysis Kaplan-Meier Estimates Number at Risk 40 39 38 37 36 35 32 31 30 29 28 27 26 25 24 22 21 19 18 17 15 13 12 11 8 7 6 5 Time 6 10 11 14 16 18 22 24 25 27 29 30 32 35 36 37 38 39 40 45 46 47 48 54 68 69 72 76 Number Failed 1 1 1 1 1 3 1 1 1 1 1 1 1 1 2 1 2 1 1 2 2 1 1 1 1 1 1 1 Survival Probability 0.00000 0. .0781625 0.44201 0.18750 0.061977 0.77503 0.645592 0.70417 0.52503 0.75000 0.295828 0.75182 0.0660256 0.65000 0.82015 0.27486 0.015435 0.271804 0.157987 0.882459 1.97500 0.0789581 0.0416458 0.115810 0.676041 0.0786607 0.0474342 0.90000 0.84201 0.448182 0.0% Normal CI Lower Upper 0. Temp100 Comparison of Survival Curves Test Statistics Method Log-Rank Wilcoxon Chi-Square 7.586626 0.000 110 Copyright © 2003–2005 Minitab Inc.97749 0.00000 0.7152 13.00000 0.60417 0. All rights reserved.41337 0.15625 0.557987 0.67500 0.0740566 0.62500 0.0740566 0.0786607 0.136626 0.529852 0.86337 0.088102 0.0724569 0.037642 0.615810 0.37500 0.0559017 95.25000 0.88419 0.370245 0.502188 0.0605154 0.92500 0.224972 0.

111 .9000 of the windings survive past 14 months. Are the survival curves for Temp80 and Temp100 significantly different? In the Test Statistics table. Output − Nonparametric Distribution Analysis The default output for Nonparametric Distribution Analysis depends on the estimation method you choose: Data Right−censored Arbitrarily−censored Kaplan−Meier estimates • • • Censoring information Characteristics of the variable. the small p-values (0. All rights reserved. 0. while at 100° C. In this case. 0. at 80° C.Distribution Analysis Graph window output Interpreting the results The estimated median failure time forTemp80 is 55 months and 38 months for Temp100.9000 of the windings survive past 31 months. Q1.005 and 0. including the mean. So the increase in temperature decreased the median failure time by approximately 17 months. a p-value < α indicates that the survival curves are significantly different. median. its standard error and 95% confidence intervals.95% confidence intervals Default estimation method Kaplan−Meier Turnbull Optional estimation method Actuarial Actuarial Copyright © 2003–2005 Minitab Inc.Standard error . The survival estimates are displayed in the Kaplan-Meier Estimates table.000) suggest that a change of 20° C plays a significant role in the breakdown of engine windings. and Q3 Kaplan-Meier estimates of survival probabilities and their . interquartile range. For example.

Reliability and Survivial Analysis

For example,

Distribution Analysis: Temp80
Variable: Temp80 Censoring Information Uncensored value Right censored value Count 37 13

Censoring value: Cens80 = 0

Nonparametric Estimates

Characteristics of Variable Standard Error 2.20686 95.0% Normal CI Lower Upper 51.3746 60.0254

Mean(MTTF) 55.7

Median = 55 IQR = * Q1 = 48

Q3 = *

Kaplan-Meier Estimates Number at Risk 50 49 48 46 45 44 43 42 41 40 39 38 35 34 33 29 28 27 26 25 24 22 21 20 19 18 17 16 13

Time 23 24 27 31 34 35 37 40 41 45 46 48 49 50 51 52 53 54 55 56 58 59 60 61 62 64 66 67 74

Number Failed 1 1 2 1 1 1 1 1 1 1 1 3 1 1 4 1 1 1 1 1 2 1 1 1 1 1 1 2 1

Survival Probability 0.980000 0.960000 0.920000 0.900000 0.880000 0.860000 0.840000 0.820000 0.800000 0.780000 0.760000 0.700000 0.680000 0.660000 0.580000 0.560000 0.540000 0.520000 0.500000 0.480000 0.440000 0.420000 0.400000 0.380000 0.360000 0.340000 0.320000 0.280000 0.258462

Standard Error 0.0197990 0.0277128 0.0383667 0.0424264 0.0459565 0.0490714 0.0518459 0.0543323 0.0565685 0.0585833 0.0603987 0.0648074 0.0659697 0.0669925 0.0697997 0.0701997 0.0704840 0.0706541 0.0707107 0.0706541 0.0701997 0.0697997 0.0692820 0.0686440 0.0678823 0.0669925 0.0659697 0.0634980 0.0621592

95.0% Normal CI Lower Upper 0.941195 1.00000 0.905684 1.00000 0.844803 0.99520 0.816846 0.98315 0.789927 0.97007 0.763822 0.95618 0.738384 0.94162 0.713511 0.92649 0.689128 0.91087 0.665179 0.89482 0.641621 0.87838 0.572980 0.82702 0.550702 0.80930 0.528697 0.79130 0.443195 0.71680 0.422411 0.69759 0.401854 0.67815 0.381521 0.65848 0.361410 0.63859 0.341521 0.61848 0.302411 0.57759 0.283195 0.55680 0.264210 0.53579 0.245460 0.51454 0.226953 0.49305 0.208697 0.47130 0.190702 0.44930 0.155546 0.40445 0.136632 0.38029

Turnbull estimates • • • Censoring information Turnbull estimates of the probability of failure and their standard errors Turnbull estimates of the survival probabilities and their standard errors and 95% confidence intervals

For example,

112

Copyright © 2003–2005 Minitab Inc. All rights reserved.

Distribution Analysis

Distribution Analysis, Start = Start and End = End
Variable Start: Start Frequency: Freq Censoring Information Right censored value Interval censored value Left censored value End: End

Count 71 694 8

Turnbull Estimates Interval Lower Upper * 10000 10000 20000 20000 30000 30000 40000 40000 50000 50000 60000 60000 70000 70000 80000 80000 90000 90000 * Probability of Failure 0.010349 0.012937 0.018111 0.032342 0.047865 0.112549 0.187581 0.298836 0.187581 0.091850 Standard Error 0.0036400 0.0040644 0.0047964 0.0063628 0.0076784 0.0113672 0.0140409 0.0164640 0.0140409 *

Time 10000 20000 30000 40000 50000 60000 70000 80000 90000

Survival Probability 0.989651 0.976714 0.958603 0.926261 0.878396 0.765847 0.578266 0.279431 0.091850

Standard Error 0.0036400 0.0054243 0.0071650 0.0093999 0.0117552 0.0152311 0.0177621 0.0161393 0.0103879

95.0% Normal CI Lower Upper 0.982516 0.996785 0.966083 0.987345 0.944560 0.972646 0.907838 0.944685 0.855356 0.901436 0.735995 0.795700 0.543453 0.613079 0.247798 0.311063 0.071490 0.112210

Actuarial survival estimates Instead of the default Kaplan-Meier or Turnbull survival estimates, you can request Actuarial estimates in the Estimate subdialog box. • • • Median residual lifetimes Conditional probabilities of failure Survival probabilities

With Nonparametric Distribution Analysis-Right Censoring, you can request specific time intervals. In this example, we requested equally spaced time intervals from 0-110, in increments of 20:

Distribution Analysis: Temp80
Variable: Temp80 Censoring Information Uncensored value Right censored value Count 37 13

Censoring value: Cens80 = 0

Nonparametric Estimates

Copyright © 2003–2005 Minitab Inc. All rights reserved.

113

Reliability and Survivial Analysis

Characteristics of Variable Standard Error 3.36718 95.0% Normal CI Lower Upper 49.5909 62.7900

Median 56.1905

Additional Time from Time T until 50% of Running Units Fail Proportion of Running Units 1.00 0.84

Time T 20 40

Additional Time 36.1905 20.0000

Standard Error 3.36718 3.08607

95.0% Normal CI Lower Upper 29.5909 42.7900 13.9514 26.0486

Actuarial Table Conditional Probability of Failure 0.000000 0.160000 0.500000 0.421053 0.000000 0.000000

Interval Lower Upper 0 20 20 40 40 60 60 80 80 100 100 120

Number Entering 50 50 42 21 9 3

Number Failed 0 8 21 8 0 0

Number Censored 0 0 0 4 6 3

Standard Error 0.000000 0.051846 0.077152 0.113269 0.000000 0.000000

Time 20 40 60 80 100 120 Note

Survival Probability 1.00000 0.84000 0.42000 0.24316 0.24316 0.24316

Standard Error 0.0000000 0.0518459 0.0697997 0.0624194 0.0624194 0.0624194

95.0% Normal CI Lower Upper 1.00000 1.00000 0.73838 0.94162 0.28320 0.55680 0.12082 0.36550 0.12082 0.36550 0.12082 0.36550

From the Results subdialog box, you can request additional output and display Session window output for each failure mode.

114

Copyright © 2003–2005 Minitab Inc. All rights reserved.

For example.Growth Curves Growth Curves Growth Curve Overview Use growth curves to analyze life data from a repairable system. Typically. By default. See Using cost or frequency columns for more information. data can also be measures such as distance from a reference point or the length of a crack. etc. Type of observation Exact data Interval data Description You know exactly when each failure occurred Example The engine failed at exactly 500 days. Frequency (or Cost) − the total cost of repairs or total frequency of failures at a particular time. Use nonparametric and parametric growth curves to determine whether a trend exists in times between successive failures of a repairable system. 115 . • Copyright © 2003–2005 Minitab Inc. see Growth curves − exact data. For example. on the type of data you have: • • • When your data consist of exact failure times. In other words. one representing failure times. you may enter the following information: • • • Time − the failure or retirement time of each sample. Remember to include a system column in this case. as in the case of identical manufacturing processes. You do not need a system column when you have only one system. and then particular times failed again sometime between 675 and 725 days How you set up your worksheet depends. such as: • • • • Setting maintenance schedules Making provisions for spare parts Assuring suitable performance Forecasting repair costs Data . was repaired. the column will contain two distinct values. to determine whether system failures are becoming more frequent. that is. System − identifies each system within a sample. then failed at 825 days You only know that each failure The engine failed sometime between 475 occurred between two and 500 days. The data points represent the time of each failure without taking into account the repair time. However. This table describes the types of observations you can have. automotive engines are usually repaired many times before being replaced. then breaks down again. one representing retirement times. Note Growth curves . System repair data usually consist of successive failure (or repair) times. an automobile breaks down. the lower value indicates the retirement time for a system. was repaired. in part. or remaining constant. All rights reserved. see Growth curves − interval data. Retirement − indicates whether the data in each corresponding row is a failure time or a retirement time. the analysis is done as if the system instantaneously works again after failing. see Growth curves − interval data or Growth curves − grouped interval data. When your data consist of failures within intervals and you are using parametric growth curves. A retirement column is not necessary if all of your data are failure truncated or time truncated.Growth Curves The data you gather for parametric and nonparametric growth curves are the failure times for a repairable system. is repaired and put back into service. A repairable system is one in which the parts are repaired instead of being replaced when they fail. Use this information to make decisions concerning the future operation of your system. When your data consist of failures within intervals and you are using nonparametric growth curves.exact data To use growth curves with exact data. Stack the system failure time data together if you believe that the rate of failures are identical. You cannot use a cost column with Parametric Growth Curve. • • Use nonparametric growth curves to estimate growth curves of the mean cost of maintaining the system or the mean number of repairs over time without making assumptions about the distribution of the cost or number of repairs. less frequent. Use parametric growth curves to estimate growth curves of the mean number of repairs and ROCOF over time using a power-law process or a homogeneous Poisson process.

You do not need a system column when you have only one system. with different colors and symbols to help you compare reliability growth between samples. system 1 failed at the first and fifth hour and is retired at the ninth hour. you can use separate columns for each sample. Frequency (or Cost) − the total cost of repairs or total frequency of failures at a particular time. All the samples display on a single plot. A failuretruncated system is retired once a certain number of failures occur. Use retirement columns if one group is time-truncated and another is failure-truncated. or date/time. A time-truncated system is retired after a specified period of time. the retirement value applies to every sample in each analysis. In the example above. Growth curves . see Data − Growth Curves. then set up a column of grouping indicators.Reliability and Survivial Analysis This data set illustrates exact data. See Using cost or frequency columns for more information.interval data Use a paired response of start and end times when your data are interval failure/retirement times. For an illustration. see Stacked vs. although pairs of columns from different samples can have different lengths. • Data with a retirement column: Enter a column of retirement indicators. When you have more than one sample. Alternatively. 116 Copyright © 2003–2005 Minitab Inc. system 1 was retired following the failure at 9 hours. Each sample is analyzed independently and results in one growth curve. • Time truncated data: Enter a column of failure times. End − the time at the end of the interval. text. See Using time and retirement columns for more information about the relationship between your time column and retirement times. Time 1 5 9 4 7 10 8 9 11 Note • System 1 1 1 2 2 2 3 3 3 Retirement 1 1 0 1 1 0 1 1 0 Frequency (or Cost) 2 4 0 2 4 0 2 4 0 Retirement times must have a value of 0 in the corresponding frequency column. If you do not specify which value indicates the retirement value in the Retirement subdialog box. Unstacked data. In a failure-truncated system. The retirement time for a system is the largest value in the variables (or time) column for that system. The retirement time is the largest failure time. System − identifies each system within a sample. while the higher one indicates a failure/repair time. system 1 was retired at 9 hours. For general information on repairable systems data. using Start and End columns: • • • • Start − the time at the start of the interval. . the system is retired immediately upon the last failure. the largest time is not a failure time. In a timetruncated system. If the systems in the example above are time-truncated and no retirement column is given. If the systems in the example above are failure-truncated. All rights reserved. Minitab assumes the lower of the two values indicates the retirement value. you can stack all the samples in one column. The columns for each sample must be the same length. If specified. Enter your data in table form. Retirement indicators can be numeric. Repairable systems data can consist of: Failure truncated data: Enter a column of failure times.

Alternatively. Unstacked data. Cost − the total cost of repairs in a particular interval. Unstacked data. Alternatively. Start 0 1 9 0 5 10 8 5 11 End 1 5 * 4 7 * 9 8 * System 1 1 1 2 2 2 3 3 3 Frequency (or Cost) 2 4 0 2 4 0 4 2 0 In this example. Copyright © 2003–2005 Minitab Inc. then set up a column of grouping indicators. All the samples display on a single plot. For an illustration. Note This data set illustrates tabled data with a cost column. • • Ten systems are running at 0 hours. All rights reserved. Enter your data in table form. see Stacked vs. then set up a column of grouping indicators. When you have more than one sample. see Stacked vs. you can use separate columns for each sample. you can stack all the samples in one column. you can use separate columns for each sample. Growth curves .Growth Curves This data set illustrates interval data. see Data − Growth Curves. For an illustration. NSystems − the number of systems running at the beginning of each interval. instead of a system column. 117 . you have a column containing the number of systems. as well as the use of a frequency column. using Start and End columns: • • • • Start − the time at the start of the interval. • • Two failures occurred between 0 and 1 hours for system 1 Observation ceased on system 1 at 9 hours When you have more than one sample. you can stack all the samples in one column. 8 repairs were made to these systems.grouped interval data Grouped interval data works like interval data except that. Between 0 and 5 hours of operation. End − the time at the end of the interval. For general information on repairable systems data. with different colors and symbols to help you compare reliability growth between samples. You must have intervals that do not overlap. Four systems are left running after the final observation time of 15 hours. Start 0 5 10 15 End 5 10 15 * NSystems 10 8 6 4 Cost/Total NRepairs 8 10 6 0 In this example. Each sample is analyzed independently and results in one growth curve.

Freq. Using Cost or Frequency Columns In a parametric analysis. Minitab assumes that your data are from identical processes and provides a pooled growth curve estimate. this column determines whether the time is a failure or a retirement time. Minitab will also compare across growth curves modeling the different processes. you may need a frequency column because of rounding error that occurred during data collection. In this case. Total frequency of failures (or number of repairs) at the time listed in the time column Cost of each repair If you do not provide a cost or frequency column. with different colors and lines to help you compare reliability growth between samples. it is not theoretically possible to have multiple failures at any one instant for a given system. and at least one of those columns corresponds to systems that are only time truncated. If a single response column represents more than one system.Reliability and Survivial Analysis Each sample is analyzed independently and results in one growth curve. . This time is not a failure time for that system. If you do not provide a retirement column and you have only failure-truncated systems. If you provide a By variable or two or more columns (exact data) or pairs of columns (interval data). you can use the column representing the cost/total number of repairs in different ways: • • Note Using Time and Retirement Columns The relationship between the time column and the retirement column is summarized below. If you have a time-truncated system. then the largest time for each system is the retirement time. • • • • If you provide a retirement column. In a nonparametric analysis. Minitab tests for equal shapes or scales across these systems. Variables/Start variables: Enter columns (one column per sample) containing the start times. However. The columns must contain positive integers (exact data) or nonnegative integers (interval data). you must use a System ID column. With interval data. All the samples display on a single plot. This results in an additional test for equal shape parameters. System ID (optional): Enter columns (one column per sample) to identify the systems. The retirement value will apply to every retirement column. then you must specify the retirement value. All rights reserved. If you provide multiple retirement columns. See Test for equal shapes or scales across growth curves. Parametric Growth Curve Parametric Growth Curve Stat > Reliability/Survival > Parametric Growth Curve Use to perform a parametric analysis on a repairable system. Dialog box items Data are exact failure/retirement times Choose if you have exact data. By variable: Check if all of the samples are stacked into one column and enter a column of grouping indicators. Minitab assumes a cost or frequency of 1 for all failures. For general information on repairable systems data. the frequency column indicates the number of failures that have occurred within the corresponding intervals. End variables: If you have interval repair times. then the retirement time for each system is the largest time for that system. These times are failure times. 118 Copyright © 2003–2005 Minitab Inc. the frequency column represents the total frequency of failures (or number of repairs) at the time listed in the time column. • • If data are exact. See Test for equal shapes or scales. Data are interval failure/retirement times Choose if you have interval data. Each system must have a retirement time. If you have a column of data from more than one system. enter columns (one column per sample) of end times. columns (optional): Enter columns (one column per sample) of frequency data. Use either a power-law process or a homogeneous Poisson process to estimate the mean number of failures or the ROCOF over time. see Data − Growth Curves.

enter the column containing the start time of each failure interval. choose System ID and enter one column for each sample to identify the systems within the sample. Then it uses the resulting estimates to provide a likelihood ratio test for: • • • • Equal shapes across groups − if you do not provide a known shape Equal scales across groups − if you provide a known shape Equal MTBFs across groups − if the shape is set at 1 H0: all of the shapes (or scales or MTBFs) are equal The hypotheses for these tests are: Copyright © 2003–2005 Minitab Inc. Choose Data are interval (failure/retirement) times. use any dialog box options. 119 . Test for Equal Shapes or Scales Across Growth Curves When your failure/repair data are from more than one growth curve. Minitab provides: • A test for equal shapes − if you do not provide a known shape • A for equality of the scale parameter − if you provide a known shape • A test for equal MTBFs − if the shape is set at 1 Each of these tests uses the Bartlett's modified likelihood ratio test whenever possible. If your data are from more than one system. If you like. Choose your model based on whether or not a trend exists in the failure/repair rate. Parametric Growth Curve Models The first step in the analysis of a repairable system is to check for a trend in the rate of failures or repairs. If you reject the null hypothesis and conclude that the shapes (or scales or MTBFs) are unequal. enter the column containing the failure/retirement times. Minitab assumes that all systems within a single data column are from identical processes. To use parametric growth curves when data are interval failure/retirement times 1 2 3 4 5 6 Choose Stat > Reliability/Survival > Parametric Growth Curve. In Variables/Start variables. Minitab first performs an analysis on each growth curve. The repair rate for a power-law process is a function of time. Minitab provides two types of models for estimating parametric growth curves: • Power-law process − Use to model failure/repair times that have an increasing. All rights reserved. or constant rate. decreasing. you cannot make this assumption. Note These tests are not available for interval data. the power-law model is also referred to as the AMSAA model. If your data are from more than one system. The hypotheses for these tests are: • • H0: all of the shapes (or scales or MTBFs) are equal H1: at least one of the shapes (or scales or MTBFs) is different When estimating a parametric growth curve. you should analyze the data from different systems separately. then click OK. choose System ID and enter one column for each sample to identify the systems within the sample.Growth Curves To use parametric growth curves when data are exact failure/retirement times 1 2 3 4 Choose Stat > Reliability/Survival > Parametric Growth Curve. In Variables/Start variables. enter the column containing the corresponding end time of each failure interval. the power-law model is also known as the Duane model. With the default (maximum likelihood) estimation method. then click OK. • Poisson process − Use to model failure/repair times that remain stable over time. With the least squares estimation method. Test for Equal Shapes or Scales When your failure/repair data are from more than one system. use any dialog box options. If you like. In this case. In End variables. You must have the same number of end variable columns as you have start variable columns.

All rights reserved. While the Anderson-Darling test is useful if you suspect the existence of a cyclic or other non-monotonic trend. Note Simulation studies have shown that a fairly large difference in p-values between TTT-based tests (including the Anderson-Darling test) and the pooled tests may indicate heterogeneity between systems. MIL-hdbk-189 (Pooled). Regardless of the model you choose. there is insufficient evidence to reject the homogeneous Poisson process model. and Anderson-Darling. Laplace's (Pooled). Use the tests for trend to determine whether a homogeneous Poisson process or a nonhomogeneous Poisson process is the appropriate model. The other tests will generally only detect monotonic trends. Thus. no trend. 120 Copyright © 2003–2005 Minitab Inc.Reliability and Survivial Analysis • H1: at least one of the shapes (or scales or MTBFs) is different These tests are not available for interval data. Trends can be: • • monotonic − times between failures are getting either consistently longer (decreasing trend) or consistently shorter (increasing trend) non-monotonic − times between failures alternate between increasing and decreasing trend (cyclic) or have a decreasing trend. Minitab provides only the MIL-Hdbk-189 test. the hypotheses for the tests for trend test are generally: H0: No trend in data (homogeneous Poisson process) H1:Trend in data (nonhomogeneous Poisson process) • • If you reject the null hypothesis. Comparison of trend tests Minitab provides five trend tests for data with multiple systems: MIL-hdbk-189 (TTT-based). You may need to analyze the data separately for each system. These tests behave differently under the following two circumstances: 1 2 the data follow a non-monotonic trend the data are from heterogeneous systems Monotonic and non-monotonic trends There is a trend in the pattern of times between failure if the times change in a systematic way. Minitab provides three trend tests: • • • • • • • • MIL-Hdbk-189 (The military handbook test) Laplace Anderson-Darling MIL-Hdbk-189 (pooled) MIL-Hdbk-189 (TTT-based) Laplace (pooled) Laplace (TTT-based) Anderson-Darling With exact data from multiple systems. The null hypothesis differs slightly depending on which test you are using. The pooled Laplace and military handbook tests reduce to their respective TTT-based tests when there is only one system. Although the power-law process may still be appropriate. . rejecting the null hypothesis means that you can definitely conclude there is a trend in your data. If you fail to reject the null hypothesis. Note Trend tests The tests for trend are sometimes referred to as goodness of fit tests. the homogeneous Poisson process is a simpler model and therefore a better choice. the other tests are more powerful in the case of a monotonic trend. and then increasing trend (bathtub) The Anderson-Darling test will reject the null hypothesis in the presence of both monotonic and non-monotonic trends. Note With exact data. Minitab provides five trend tests: With interval data. Laplace's (TTT-based). See Comparisons of trend tests for more information about which test is most relevant for your data. Homogeneous and heterogeneous systems The null hypothesis of no trend differs slightly for the different tests: • The null hypothesis for the pooled tests (MIL-hdbk-189 and Laplace's) is that the data come from a homogeneous Poisson processes (HPP) with a possibly different MTBF for each system. you can conclude that there is some trend in your data and you should model your data with a nonhomogeneous Poisson process such as the power-law process.

All retirement columns must have exactly two distinct values unless you specify a retirement value. the second retirement column is paired with the second data column. the probability of at least one failure in the next 7 days is: Parametric Growth Curve − Retirement Stat > Reliability/Survival > Parametric Growth Curve > Retirement Use to specify the retirement information. Time truncated systems: Choose if you have time-truncated data only. Failure truncated systems: Choose if you have failure-truncated data only. rejecting the null hypothesis could mean that either there is a trend in your data or your data come from heterogeneous systems. that is. the system is retired immediately upon the last failure. The estimated mean cumulative function ( where • t = the time since the start of the test • • = the estimated shape parameter = the estimated scale parameter )= The probability that at least one failure will occur between now (t) and the next t1 days is: where X is the number of failures in the time interval (t. Dialog box items Retirement time at largest time for system Choose if the largest time for each system is the retirement time for the system. You use Parametric Growth Curve to estimate the current value of the MCF and obtain the following estimates of shape and scale respectively: • • = 0. All rights reserved. assume you are analyzing machinery breakdown and want to find the probability of at least one breakdown in the next seven days. Therefore. Thus. HPP (possibly different MTBFs) monotonic trend MIL-hdbk-189 (TTT-based) HPP (equal MTBFs) Laplace's (Pooled) HPP (possibly different MTBFs) Laplace's (TTT-based) HPP (equal MTBFs) Anderson-Darling HPP (possibly different MTBFs) monotonic trend or monotonic trend systems are heterogeneous monotonic trend or monotonic trend or systems are non-monotonic heterogeneous trend or systems are heterogeneous See [12] for more information concerning these tests. you should use TTT-based tests only when you are confident that your systems are homogeneous. Copyright © 2003–2005 Minitab Inc. the largest time for each system is not a failure time. MIL-hdbk-189 (Pooled) Null Hypothesis Rejecting H0 means. and so on. Retirement columns: Enter the retirement columns in the box. For example. t+t1]..5 = 3. Laplace's. and Anderson-Darling) is that the data come from a homogeneous Posson process (HPP) with the same MTBF for each system. The table below summarizes the different null hypotheses associated with the trend tests. that is. Retirement time defined by retirement columns Choose to define the retirement time when you have both failuretruncated and time-truncated systems.5 If the test has already been running for t = 8 days. 121 . See Using time and retirement columns for more information. The first retirement column is paired with the first data column. Using Parametric Growth Curve to find the probability of failure You can use the results from Parametric Growth Curve to estimate the probability of failure within a given time frame.Growth Curves • The null hypothesis for the TTT-based tests (MIL-hdbk-189..

. The differences are shown in the table below: Method Maximum likelihood Maximum likelihood Maximum likelihood Least squares Least squares Least squares Shape Unknown Known. When estimating the shape and scale parameters. Estimate shape parameter: Choose to estimate the shape parameter from the data. and least squares. See Parametric growth curve models. Parametric Growth Curve − Estimate Stat > Reliability/Survival > Parametric Growth Curve > Estimate Use to specify one of three estimation methods and either a power-law process or Poisson process model. Minitab displays output in the Session window based on the chosen estimation method and whether or not the shape parameter is known. and least squares. conditional maximum likelihood. and least squares. See Parametric growth curve models. See Estimation methods for parametric growth curves.0%. Minitab will make this number as large as possible. = 1 Known. the lowest value in the retirement column is the retirement value. Estimation methods for parametric growth curves Minitab estimates the shape and scale using one of three estimation methods: maximum likelihood. Dialog box items Estimation method: Choose one of three estimation methods: maximum likelihood. Note You cannot use a retirement column when you have interval data. Text values must be contained in double quotes.Reliability and Survivial Analysis Retirement value: Enter a value indicating which value in your retirement column represents a retired system. The table below summarizes the features available with each estimation method. not 1 Parameter estimates Shape and scale MTBF Scale Shape and scale Shape and scale MTBF Scale Standard error and confidence intervals Yes Yes Yes For shape only No No No Conditional maximum likelihood Unknown See Comparison of growth curve procedures for a discussion of the advantages and disadvantages of each method. even when data are from multiple systems Comparison of estimates between growth curves Available for interval data Available when the shape is specified No Yes Yes No Yes Yes Yes CMLE Yes No No Yes Yes No No LSYX No No No Yes No Yes Yes 122 Copyright © 2003–2005 Minitab Inc. Set shape parameter: Choose to specify the shape value and enter a positive numeric constant. Poisson process: Choose to model the data using a Poisson process. Confidence intervals: Choose to use two-sided confidence intervals (the default) or just an upper or lower confidence bound. Minitab calculates the maximum likelihood and conditional maximum likelihood estimates. not 1 Unknown Known. MLE Exact standard error and confidence interval for the shape Asymptotic standard error and confidence interval for the shape Asymptotic standard error and confidence interval for the scale Non-iterative solutions. = 1 Known. conditional maximum likelihood. Power-Law process Choose to model the data using a Power-Law process. You can also enter a confidence level for Minitab to use for all confidence intervals. All rights reserved. Comparison of growth curve procedures Minitab provides three estimation methods for the parameter estimates: maximum likelihood. The default is 95. The log-likelihood is a measure of the fit of the distribution. conditional maximum likelihood. If you do not enter a value. By maximizing the likelihood function. Confidence level: Enter a confidence level for all of the confidence intervals.

differs depending on your model. Maximum X scale: Enter a value for the maximum x-axis scale. The resulting graph can be straight or a curve that is either concave up or down. Dialog box items Event Plot Check to display an event plot. and deteriorating systems. Display number of failures on event plot: Check to display the frequency of failures on the event plot. 123 . All rights reserved. however. Duane plot: Check to display a Duane plot. The plot provides information about the pattern of system failures: • • • A straight line pattern indicates that system failures are remaining constant over time − your system is stable A concave down pattern indicates that the time between failures is increasing over time − your system reliability is improving A concave up pattern indicates that the time between failures is decreasing over time − your system reliability is deteriorating Below are examples of mean cumulative function plots for improving. For a homogeneous Poisson process. which is a plot of the empirical mean cumulative function. or total-time-on-test plot. One plot per system: Choose to create separate plots for each system. Minitab estimates failure times by evenly distributing the number of occurrences in each interval and plotting the appropriate points. the rate of system failures can increase. which is a plot of the mean cumulative function based on the estimated shape and scale. The plot consists of: • The Nelson-Aalen plot. This option does not apply to the event plot. Mean Cumulative Function and Nelson-Aalen Plot Use the mean cumulative function and Nelson − Aalen plot to determine whether your system is improving.Growth Curves Parametric Growth Curve − Graphs Stat > Reliability/Survival > Parametric Growth Curve > Graphs Use to create an event plot. Show graphs of different variables/by levels or systems: Choose to display the graphs overlaid on the same page or on separate pages. X axis label: Type the desired text to replace the default x-axis label with your own label. This option does not apply to the TTT plot. • Because the Nelson-Aalen plot does not depend on the model. Duane plot. This option does not apply to the TTT plot. stable. decrease. Mean cumulative function (MCF) and Nelson-Aalen plot: Check to display the estimated mean cumulative function overlaid on the Nelson-Aalen plot. The mean cumulative function plot. The plot points do not assume a particular model. mean cumulative function and Nelson-Aalen plot. Total time on test (TTT) plot Check to display a total-time-on-test plot. the plot points are the same regardless of which estimation method and model type you chose. the failure rate is constant. Minitab plots a 95% confidence interval. When you have interval data. resulting in a straight line. deteriorating. Display confidence intervals on MCF plot: Check to display confidence intervals on the MCF plot. This option does not apply to the event plot. You can change the confidence level in the Estimate subdialog box. or remain constant. X axis display (except TTT plot): Minimum X scale: Enter a value for the minimum x-axis scale. The mean cumulative function plot. Copyright © 2003–2005 Minitab Inc. or staying constant. For a power-law process. This option does not apply to the TTT plot. Display (except event plot): An average plot over all systems: Choose to create plots which are averaged over (or pooled across) all systems. By default.

124 Copyright © 2003–2005 Minitab Inc. decreasing. stable. The plot consists of: • • • Horizontal lines. and deteriorating systems. . which represent the lifetime of each system Cross (X) points. or remaining constant. Below are examples of event plots for improving. which represent the failure and retirement times of each system Cost values or frequencies (optional).Reliability and Survivial Analysis Event plot A plot of events (failures and retirements) for all systems. All rights reserved. which represent the cost or frequency of failure at the cross points Use the event plot to visually determine whether successive failures are increasing.

Use a Duane plot to: • • Assess whether your data follow a power-law process or a homogeneous Poisson process Determine if your system is improving. See Methods and Formulas for the formula for the fitted line. and no slope (a horizontal line) shows a stable system. the event plot is constructed using the midpoint of each interval. A negative slope shows reliability improvement. The Duane plot should be roughly linear if the power-law process or homogeneous Poisson process is appropriate. Duane plot A scatterplot of the cumulative number of failures at a particular time divided by the time (cumulative failure rate) versus time. All rights reserved.Growth Curves Note For interval data. deteriorating. or remaining stable The fitted line on the Duane plot is the best fitted line when the assumption of the power-law process is valid and the shape and scale are estimated using the least squares method. Copyright © 2003–2005 Minitab Inc. Below are examples of Duane plots for improving. stable. 125 . and deteriorating systems. a positive slope shows reliability deterioration.

stable. 126 Copyright © 2003–2005 Minitab Inc. The TTT plot provides a graphical goodness-of-fit test for the power-law process. • • A power-law process is appropriate if the TTT plot lies close to the diagonal or is a curve that is either concave up or concave down If there is no pattern. the power-law process is inadequate Below are examples of total time on test plots for improving. All rights reserved. or a curve that shifts between being concave up and concave down.Reliability and Survivial Analysis Total Time on Test Plot Use a total-time-on-test (TTT) plot to visualize how well your model fits the data. and deteriorating systems. .

the last shape value listed is used for the remaining variables. Parametric Growth Curve − Options Stat > Reliability/Survival > Parametric Growth Curve > Options Use to specify starting estimates for the shape and the maximum number of iterations when available. Copyright © 2003–2005 Minitab Inc. Parametric Growth Curve − Storage Stat > Reliability/Survival > Parametric Growth Curve > Storage Use to store various characteristics of the parametric growth curve. Mean cumulative function (MCF): Store the mean cumulative function evaluated at the corresponding time (exact data) or end time (interval data). You can provide as many starting estimates as there are response columns. Minitab stores the 95% confidence limits. You can change the confidence level in the Estimate subdialog box. Dialog box items Show log-likelihood for each iteration of algorithm Check to display the value of the shape. and the loglikelihood at each iteration when available. Dialog box items Enter number of levels in by variable: If all of the samples are stacked in one column. 127 . If you provide fewer starting estimates than there are response columns. enter the number of levels the column of grouping indicators contains. Dialog box items Use starting estimate for shape: Enter starting estimates for the shape. By default. scale. All rights reserved. Confidence limits for MCF: Check to store the confidence limits for the mean cumulative function.Growth Curves Parametric Growth Curve − Results Stat > Reliability/Survival > Parametric Growth Curve > Results Use to control the display of the Session window output. The default number of iterations is 20. Maximum number of iterations: Enter a positive integer to specify the number of iterations.

Session window output Parametric Growth Curve: Days System: ID Model: Power-Law Process Estimation Method: Maximum Likelihood Parameter Estimates Standard Error 0.229 458.130 Test for Equal Shape Parameters Bartlett's Modified Likelihood Ratio Chi-Square Test Statistic P-Value DF 23.73 33. All rights reserved. .28239 4.406 18. enter Days. Minitab stores the 95% confidence limits. Choose Stat > Reliability/Survival > Parametric Growth Curve.59 0. enter ID. Click OK.88436 385.425 23 Trend Tests MIL-Hdbk-189 TTT-based Pooled 34. Confidence limits for ROCOF: Check to store the confidence limits for the ROCOF.13 5. Example of Parametric Growth Curve You want to estimate the replacement rate of a certain valve on a fleet of 65 diesel engines.00405 420.590 95% Normal CI Lower Upper 3.Reliability and Survivial Analysis Failure rate (ROCOF): Check to store the ROCOF evaluated at the corresponding time (exact data) or end time (interval data).000 0.000 128 Copyright © 2003–2005 Minitab Inc. By default. Standard error of estimates: Check to store the standard errors of the shape and scale where available.000 142 96 Laplace's TTT-based Pooled 9.20 0.MTW.73 0. Parameter estimates: Check to store estimates of the shape and scale. In System ID. Minitab stores the 95% confidence limits. You can change the confidence level in the Estimate subdialog box. 1 2 3 4 5 Open the worksheet VSEAT.62 0.000 0. By default. Confidence limits for parameters: Check to store the confidence limits for the shape and scale. In Variable/Start variables.193 Parameter Shape Scale Estimate 4.000 Test Statistic P-Value DF Anderson-Darling 50. You can change the confidence level in the Estimate subdialog box.

indicating that the failure rate is increasing.282. You can be 95% confident that the interval (3. 4.Growth Curves Graph window output Interpreting the results The estimate of the shape (4.884) contains the true shape.004) is greater than 1. 129 . All rights reserved. Copyright © 2003–2005 Minitab Inc.

you must use a System ID column. In Variables/Start variables. This means that there is enough evidence to reject the null hypothesis that there is no trend in your data. To use nonparametric growth curves when data are exact (failure/retirement) times 1 2 3 4 Choose Stat > Reliability/Survival > Nonparametric Growth Curve. enter the column containing the end time of each failure interval. You can conclude that the increasing trend is significant. Minitab assumes that all data are from the same system. enter the column containing the start time of each failure interval. Nonparametric Growth Curve Nonparametric Growth Curve Stat > Reliability/Survival > Nonparametric Growth Curve Use to perform a nonparametric analysis on a repairable system when you want to estimate a growth curve of the mean cost of maintaining a system or the mean number of repairs over time without assuming a distribution. Choose Data are interval (failure/retirement) times. choose System ID and enter one column for each sample to identify the systems within the sample. then enter one column for each sample to specify the number systems entering each time interval. Minitab compares the different repairable systems. If you provide two or more columns (exact data) or pairs of columns (interval data). If your data are from more than one system. The pooled estimate of the shape is valid. System Information If you do not provide any system information. To use nonparametric growth curves when data are interval (failure/retirement) times 1 2 3 4 5 Choose Stat > Reliability/Survival > Nonparametric Growth Curve. In End variables. do one of the following: • Choose System ID and enter one column for each sample to identify the systems within the sample • Choose Number of systems and enter one column for each sample to specify the number of systems entering each time interval If you like. If you like. Your data must be in interval format to use this option. System ID: Choose to identify the systems within the sample. By variable: Check if all of the samples are stacked in one column and enter a column of grouping indicators. then enter one column for each sample. If a single response column represents more than one system. enter the column containing the failure/retirement times. use any dialog box options. End variables: If you have interval repair times. use any dialog box options.Reliability and Survivial Analysis The test for equal shape parameters indicates that there is not enough evidence to say that the systems come from populations with different shapes (P-Value = 0. You must have the same number of end variable columns as you have start variable columns. Dialog box items Data are exact failure/retirement times Choose if you have exact data.000). If your data are from more than one system. The plot of the MCF versus time shows a curve that is concave up. then click OK. This plot is consistent with a shape that is greater than one. then click OK. The event plot seems to show a pattern of failures that become more frequent as time goes on. Number of systems: Choose if your data are grouped. In Variables/Start variables. enter columns (one column per sample) of end times. The tests for trend are all significant (P-Value = 0. All rights reserved. or a system that is deteriorating. 6 130 Copyright © 2003–2005 Minitab Inc.425). Data are interval failure/retirement times Choose if you have interval data. . Variables/Start variables: Enter columns (one column per sample) containing the start times.

Minitab assumes a cost of 1 for all failures. the second retirement column is paired with the second data column. the lowest value in the retirement column is the retirement value. Display confidence intervals on plots: Check to display confidence intervals on the mean cumulative function and mean cumulative difference function plots. or plot of the mean cumulative difference function. Retirement columns: Enter the retirement columns in the box. Nonparametric Growth Curve − Graphs Stat > Reliability/Survival > Nonparametric Growth Curve > Graphs Use to create an event plot. All rights reserved. The value in this column must be 0 for a retired system. The column must contain numeric values greater than or equal to 0. If you do not provide this column. plot of the mean cumulative function. X axis label: Type the desired text to replace the default x-axis label with your own label. 131 . which represent the lifetime of each system Copyright © 2003–2005 Minitab Inc. The first cost column is paired with the first data column. the second cost column is paired with the second data column. Minitab plots a 95% confidence interval. By default.Growth Curves Nonparametric Growth Curve − Retirement Stat > Reliability/Survival> Nonparametric Growth Curve > Retirement Use to specify the retirement information. Maximum X scale: Enter a value for the maximum x-axis scale. and so on. Retirement time defined by retirement columns Choose to define the retirement time. that is. Time truncated systems: Choose if you have time-truncated data only. and so on. Dialog box items Repair cost/number of repairs for systems: Enter a column containing the total cost of repairs or total frequency of repairs. Dialog box items Retirement time at largest time for system Choose if the largest time for each system is the retirement time for the system. Failure truncated systems: Choose if you have failure-truncated data only. See Using time and retirement columns for more information. the largest time for each system is not a failure time. Nonparametric Growth Curve − Cost-Freq Stat > Reliability/Survival > Nonparametric Growth Curve > Cost-Freq Use to enter the total cost of repairs or number of repairs for each row of data. You can change the confidence level in the Options subdialog box. that is. Minimum X scale: Enter a value for the minimum x-axis scale. Note You cannot use a retirement column when you have interval data. Show graphs of different variables or by levels: Choose to display the graphs overlaid on the same page or on separate pages. the system is retired immediately upon the last failure. Display cost/number of repairs on event plot: Check to display the cost or number of repairs on the event plot. Confidence intervals are only calculated when you have multiple systems. Mean cumulative function: Check to display a plot of the mean cumulative function. The first retirement column is paired with the first data column. All retirement columns must have exactly two distinct values unless you specify a retirement value. Mean cumulative difference function: Check to display a plot of the mean cumulative difference function. Dialog box items Event Plot Check to display an event plot. Event plot A plot of events (failures and retirements) for all systems. If you do not enter a value. Text values must be contained in double quotes. Retirement value: If you like. You must provide a cost/frequency column if you have grouped interval data. The plot consists of: • Horizontal lines. Event plots are always on separate pages. This plot is only available when you have multiple samples. enter a value indicating which value in your retirement column represents a retired system.

Reliability and Survivial Analysis • • Cross (X) points. When you have interval data. and deteriorating systems. stable. Below are examples of event plots for improving. or remaining constant. All rights reserved. stable. the event plot is constructed using the midpoint of each interval. The plot is a step function with steps at system failures or endpoints of intervals. decreasing. which represent the failure and retirement times of each system Cost values or frequencies (optional). The mean cumulative function plot displays the mean cumulative function versus time. The plot provides information about the pattern of system failures: • • • A straight line indicates that system failures are remaining constant over time A curve that is concave down indicates that the time between failures is increasing over time − your system reliability is improving A curve that is concave up indicates that the time between failures is decreasing over time − your system reliability is deteriorating Below are examples of mean cumulative function plots for improving. and deteriorating systems. Mean Cumulative Function Plot Use the mean cumulative function plot to visualize whether your system reliability is improving. . which represent the cost or frequency of failure at the cross points Use the event plot to visually determine whether successive failures are increasing. Note For interval data. or staying constant. 132 Copyright © 2003–2005 Minitab Inc. Minitab plots the middle point. deteriorating.

The mean cumulative difference function plot displays the mean cumulative difference function versus time.Growth Curves Mean Cumulative Difference Function Plot Use the mean cumulative difference function plot to visualize the differences in the mean cumulative function across two or more groups over time. 133 . All rights reserved. Copyright © 2003–2005 Minitab Inc. Minitab places a reference line where the mean cumulative difference function is equal to zero. This plot is only available when you have one of the following: • • • Two or more variables Two or more pairs of variables (interval data) A By variable Below is an example of a mean cumulative difference function plot. The plot is a step function with steps at system failures or endpoints of intervals.

By default. Standard error of MCF: Check to store the standard error of the mean cumulative function. All rights reserved. Times for difference in MCF: Check to store the times associated with the mean cumulative difference function. 134 Copyright © 2003–2005 Minitab Inc. Confidence intervals: Choose to use two-sided confidence intervals (the default) or just an upper or lower confidence bound. Confidence limits for MCF: Check to store the confidence limits for the mean cumulative function. Times for MCF: Check to store the times associated with the mean cumulative function. You can change the confidence level in the Options subdialog box. Dialog box items Enter number of levels in by variable: Enter the number of levels the column of grouping indicators contains. Minitab stores the 95% confidence limits.Reliability and Survivial Analysis Nonparametric Growth Curve − Options Stat > Reliability/Survival > Nonparametric Growth Curve > Options Use to specify the confidence level and type of confidence interval. . The default is 95. Standard error of difference in MCF: Check to store the standard error of the mean cumulative difference function. Nonparametric Growth Curve − Storage Stat > Reliability/Survival > Nonparametric Growth Curve > Storage Use to store various characteristics of the nonparametric growth curve. Dialog box items Confidence level: Enter a confidence level for all of the confidence intervals.0%. Difference in individual Difference in mean cumulative function (MCF): Check to store the mean cumulative difference function. Use only if all of the samples are stacked in one column. Individual curve Mean cumulative function (MCF): Check to store the mean cumulative function.

137661 0.158574 0.23407 0. choose System ID.260862 0.26653 0.95456 1.91581 2.12481 0.149098 0.21429 0.54468 0.268755 0.10744 2.280566 0.03047 3.40692 2.64286 1. Check By variable.29771 0.03084 3.89833 2.32679 2. You can change the confidence level in the Options subdialog box.46596 1.07143 0.58771 2.28571 1.07859 0.74258 1.52456 1. Minitab stores the 95% confidence limits.58008 1. The final time for each train is the final failure for that train.03047 2.132260 0.14286 0.50000 1.71429 0.443994 0.46380 2.00000 1.02872 1.49012 2.98273 2.20522 2.71429 1.14424 System 179 132 128 137 181 119 182 112 167 112 175 137 155 119 101 145 182 119 128 112 137 137 169 169 182 179 181 175 169 145 155 167 137 112 128 132 Copyright © 2003–2005 Minitab Inc.96675 1.43506 0.410559 95% Normal CI Lower Upper 0.35714 0.425594 0.85714 0.40338 1.MTW.270649 0.58426 0.93013 1.068830 0.35714 1.01081 0.78571 0.10800 3.51818 4.59671 1.240267 0.191853 0.27912 2.27405 4.70992 1.24158 2.57143 0.128060 0.83047 3.84423 0.259653 0.05735 1.128060 0.79042 1.132260 0.109664 0.57143 1.54168 1.133631 0.07143 1.274527 0.174964 0. Your data include replacement times and component type for 29 trains.10019 0.120736 0.23047 Time 33 88 250 272 287 302 317 364 367 391 402 421 431 444 462 481 498 500 500 548 552 625 635 650 657 687 687 700 708 710 710 710 719 724 724 724 Standard Error 0.80055 2. 1 2 3 4 5 6 Open the worksheet TRAIN.80165 1.65498 2.09304 2.29750 1.13970 0. Session window output Nonparametric Growth Curve: Days Results for Type = 1 System: ID Nonparametric Estimates Table of Mean Cumulative Function Mean Cumulative Function 0.40906 0.35269 2.Growth Curves Confidence limits for difference in MCF: Check to store the confidence limits for the mean cumulative difference function.65408 0.78571 1.254826 0.256120 0. enter Days.187044 0.42857 1.242226 0.93956 2.64286 0.05848 1.17686 0.157421 0.03516 2.70969 1.23940 1.77214 2.22982 1.43200 0. Under System Information. By default. All rights reserved.093522 0.266655 0.83537 1.66444 1.347216 0. In Variables/Start variables.14286 1.72120 0.46374 1.50000 0.78471 0.63047 2.149098 0.40714 2.219328 0.29613 0.86264 1. Example of Nonparametric Growth Curve You want to compare two different types of a particular brake component used on a subway train.89945 0.94990 0. then enter Type.21429 1. Choose Stat > Reliability/Survival > Nonparametric Growth Curve.92857 1.03960 0.47218 0.44718 0.170747 0. 135 .13047 2. then enter ID.158574 0.36303 0.90253 1.257586 0.42857 0. Click OK.51540 0.03849 2.28571 0.

73047 4.46667 1.349310 0.44284 0.63337 3.23047 0.263312 0.32729 2.309839 0.264435 0.54335 1.53333 0.35227 1.09468 2.13333 0.114180 0.324779 0.47298 1.83985 1.66667 1.07269 0.73333 0.535360 0.03670 0.410559 2.58420 2.67333 1.91948 1.20810 2.66667 0.32237 1.257624 0.18831 System 228 212 192 214 219 192 243 192 192 190 228 235 205 216 183 243 183 190 204 243 183 184 192 206 183 200 243 235 219 228 216 204 214 212 205 206 200 192 190 184 183 Comparisons for Days Comparison: (Type = 1) .00696 1.14538 0.33333 1.064406 0.207989 0.87143 1.257624 0.30945 1.335548 0.44435 2.341839 0.06667 1.249444 0.12325 2.342540 0.157762 0.73333 1.61370 1.20000 1. All rights reserved.00000 1.41809 0. .22025 4.59515 2.48447 0.49769 4.535360 95% Normal CI Lower Upper 0.11618 2.00786 2.74512 3.43622 0.52661 0.41289 1.91221 3.47658 1.449834 0.13333 1.20364 2.71738 1.09778 5.11630 1.02134 Time 19 22 39 54 61 91 93 119 148 173 185 187 192 194 203 205 211 242 250 264 277 293 306 369 373 382 415 416 419 419 432 434 441 447 448 448 460 461 464 503 511 Standard Error 0.274334 0.31007 2.11521 0.93801 3.02109 3.24834 1.63052 0.298009 0.41776 4.257624 0.53391 2.345034 0.40000 0.73764 1.277555 0.45229 2.86667 0.087771 0.33333 0.261052 0.06667 0.55006 0.18466 3.88664 2.52134 4.159629 0.11677 101 119 Results for Type = 2 System: ID Nonparametric Estimates Table of Mean Cumulative Function Mean Cumulative Function 0.77864 5.60000 1.92615 2.21618 2.61330 1.16295 0.38662 2.53333 1.21141 2.85023 1.33859 1.280740 0.Reliability and Survivial Analysis 730 730 3.73801 2.60000 0.48693 1.26667 1.94835 2.88706 0.66442 1.79046 0.43152 2.337430 0.01004 0.26667 0.338097 0.86652 0.40000 1.02134 5.295146 0.68186 0.315398 0.71243 0.471307 0.00507 0.340512 0.60321 0.93333 1.20000 0.61721 0.57604 0.121716 0.01238 2.347441 0.80000 0.10369 1.18465 0.25386 1.86504 0.33227 2.77210 1.23869 0.68108 3.87262 3.13434 1.55029 0.264435 0.46667 0.511478 0.07443 6.331033 0.269979 0.103280 0.18801 3.40832 3.91237 0.23962 2.269979 0.02527 2.(Type = 2) 136 Copyright © 2003–2005 Minitab Inc.319258 0.80000 1.64893 4.

93616 -0. All rights reserved.24201 -3.546276 0.321731 0.93018 -0.280486 0.02241 -1.39048 -0.15671 -0.23956 -1.285719 0.36965 -2.153496 0.41664 -3.04286 -1.12857 -0.19726 -0.64634 -0.68642 0.05238 -1.05988 Copyright © 2003–2005 Minitab Inc.10231 -0.03333 -0.21714 -0.06730 0.350677 0.274074 0.70397 -0.13729 -0.99048 -1.45438 -1.26527 -2.72166 -0.285719 0.77883 -4.95714 -1.01905 -0.54467 -4.81428 -0.66420 -3.16813 -1.578546 0.376828 0.06607 -0.03880 -0.57473 -0.59277 -3.73929 -0.09048 -1.89048 -0.Growth Curves Table of Mean Cumulative Difference Function Mean Cumulative Difference Function -0.37183 0.228047 0.63823 -0.05957 -0.32729 -1.25714 -0.77883 -4.45229 -1.76467 -0.380094 0.47024 -4.04762 -1.05714 -1.568698 0.32381 -0.308568 0.30536 0.20622 -1.185008 0.00481 0.41628 -1.92381 -0.83744 0.70733 -2.139830 0.42985 -1.064406 0.32765 -0.11658 -2.13401 -1.11469 -0.65714 -0.01969 -1.28761 -1.371724 0.05649 -1.18174 -1.599884 95% Normal CI Lower Upper -0.310335 0.532767 0.44073 -1.54957 -2.44991 -2.381917 0.35126 -3.45714 -0.277299 0.302679 0.67344 -0.568698 0.52381 -0.93801 -2.71672 -0.38663 -1.37928 -0.26070 -3.79208 -0.61660 0.46099 -0.07995 -0.359155 0.351196 0.71431 -0.10476 -1.379800 0.61424 -0.54022 -0.298435 0.345574 0.19524 -0.01216 -0.49947 -3.40608 -1.79048 -0.60116 -0.26285 -1.535496 0.14387 -2.379729 0.27459 -1.09670 -1.279427 0.57464 -2.482663 0.32138 -1.98395 -0.34033 -1.66015 -2.61532 -1.84858 -0.85714 -0.349114 0.43907 -2.25701 -1.50360 -1.111541 0.03869 -0.49132 0. 137 .529680 0.41745 -1.373582 0.11905 -1.72381 -0.087771 0.343985 0.19048 -0.02381 -1.18761 -1.17407 -1.59048 -0.350677 0.94762 -1.384013 0.183399 0.18801 -2.352358 0.21230 -2.23563 Time 19 22 33 39 54 61 88 91 93 119 148 173 185 187 192 194 203 205 211 242 250 264 272 277 287 293 302 306 317 364 367 369 373 382 391 402 415 416 419 421 431 432 434 441 444 447 448 460 461 462 464 481 498 500 503 511 548 552 625 635 650 657 Standard Error 0.73801 -1.45884 -2.30004 -1.133322 0.25330 -1.48928 -1.11429 -1.33501 -1.388808 0.15855 -0.335260 0.34482 -1.08571 -1.37849 -3.26190 -0.28208 -1.280486 0.28052 0.30706 -2.30706 -3.37849 -2.44991 -3.360677 0.587608 0.52134 -3.09352 -1.72670 -4.44492 -1.01429 -1.63456 -4.595004 0.34500 -1.54957 -2.03109 -2.11037 -0.355960 0.16420 -2.49517 -1.318886 0.41228 -1.39872 -1.13333 -0.03187 -0.67303 -4.23488 -3.66420 -3.31988 -1.279994 0.09052 -1.10952 -1.45654 0.61714 -0.72774 -0.06190 -0.09714 -1.05142 -3.53413 -1.593471 0.35715 -1.604423 0.19290 0.476966 0.64922 -0.96190 -0.289837 0.50474 -1.383618 0.64086 -0.53597 0.27355 -0.03810 -1.266399 0.49714 -1.06667 -0.74649 -0.124114 0.54336 -2.41138 -1.274074 0.40036 -1.

14022 -0.15296 -4.46856 0.11319 -1. .91456 -1.638098 0.07007 -3.82878 -1.70765 -3.79087 -0.53144 Graph window output 138 Copyright © 2003–2005 Minitab Inc.64152 -3.Reliability and Survivial Analysis 687 700 708 710 719 724 730 -3. All rights reserved.71167 -1.11319 -2.40743 -1.601644 0.99087 -2.674662 -4.674662 0.595533 0.89087 -2.24900 -4.39087 -1.592914 0.79087 0.55754 -2.586803 0.08178 -2.

All rights reserved.47218. Here. The mean cumulative function plot displays the mean cumulative function for each group. type 2 trains had. For example. Because all of the confidence intervals contain zero. indicating that the failure rate is remaining relatively constant until that point The line representing type 1 trains is slightly concave up. Each line extends to the last day of observation. Use this plot to visualize trends within and across groups. not curved. 0. the mean cumulative function is 0. up until around 450 days. indicating that the failure rate is slightly increasing The line representing type 1 trains is to the right of the line representing type 2 trains. you can conclude that: • • • The line representing type 2 trains is relatively straight. Use the mean cumulative difference function to make comparisons across groups. indicating that failures are occurring less often for type 1 trains Copyright © 2003–2005 Minitab Inc.01081 and 0. 139 .07143 at 33 days for the first type of train. system failures are occurring at a constant rate. The first failures occur slightly earlier for the type 1 trains. no significant differences exist in the mean cumulative difference function between groups at any given time. For example.06667 more failures at 19 days. From this plot. The event plot shows when the failures occurred for each system. on average. You can be 95% confident that the true mean cumulative function is between 0.Growth Curves Interpreting the results Minitab displays nonparametric estimates of the mean cumulative function and its corresponding standard error and confidence limits separately for each group.

.

normal. This model will help you understand how different factors and covariates affect the lifetime of your part or product. its levels exceed those normally found in the field. All rights reserved. The second predictor can be either a second accelerating variable or a factor. • Both regression with life data commands differ from other regression commands in Minitab in that they use different distributions and accept censored data. Responses are uncens/arbitrarily censored data: Choose if your data is uncensored or arbitrarily censored. 50th. End variables: If you have uncensored or arbitrarily censored data. inverse temperature. Life data is often incomplete or censored in some way. Accelerated life testing requires knowledge of the relationship between the accelerating variable(s) and failure times. You can choose to model your data on one of the following eight distributions: Weibull. Arrhenius. Accelerated Life Testing Stat > Reliability/Survival > Accelerated Life Testing Use accelerated life testing to investigate the relationship between failure time and one or two predictors. enter up to 10 columns (10 different samples) of end times. and nested terms. Minitab assumes the relationship is linear. Here are the steps: 1 2 3 Impose levels of the accelerating variable(s) on the units. The variable is thus called the accelerating variable. or organism. Record failure (or censoring) times. The most common application of accelerated life testing is for studies in which you impose a series of variable levels far exceeding normal field conditions to accelerate the failure process. The goal is to come up with a model that predicts failure time. Run the Accelerated Life Testing analysis. Regression with Life Data performs a regression with one or more predictors. Censored observations are those for which an exact failure time is unknown. smallest extreme value. since under normal field conditions. it can take a very long time for a unit to fail. Variables/Start variables: Enter up to 10 columns (10 different samples) containing the start times. relation plot. product. The model can include factors. Copyright © 2003–2005 Minitab Inc. part. logistic. Based on these predictions you can estimate the reliability of the system. you can find out how the units behave under normal field conditions. and loglogistic. The relation plot displays the relationship between the accelerating variable(s) and failure time by plotting percentiles for each level of the accelerating variable(s). The data obtained under the high stress conditions can then be used to extrapolate back to normal use conditions. lines are drawn at the 10th. you may only know that a product failed before a certain time. This way. columns (optional): Enter a column containing frequency data for each variable. The first predictor is an accelerating variable. Suppose you are testing how long a product lasts and you plan to end the study after a certain amount of time. and 90th percentiles. covariates. lognormal basee. Accelerating var: Enter the column containing the predictor values. exponential. The probability plot is created for each level of the accelerating variable(s) based on the fitted model (line) and based on the nonparametric model (points). or loge (power) transformation for the accelerating variable. 141 . The second predictor can be either a second accelerating variable or a factor. In order to do this. The simplest output includes a regression table. Any products that have not failed before the study ends are right-censored. Freq. Similarly.Accelerated Life Testing Accelerated Life Testing Regression with Life Data Overview Use Minitab's regression with life data commands to investigate the relationship between failure time and one or more predictors. you must have a good model of the relationship between failure time and the accelerating variable(s). which is left-censored. meaning that the part failed sometime after the present time. • Accelerated Life Testing performs a simple regression with one or two predictors that is used to model failure times for highly reliable products. interactions. Minitab uses a modified Newton-Raphson algorithm to calculate maximum likelihood estimates of the model parameters. Accelerated tests are performed to save time and money. Dialog box items Responses are uncens/right censored data: Choose if your data is uncensored or right censored. The 50th percentile is a good estimate for the time a part will last when exposed to various levels of the accelerating variable(s). The first predictor is an accelerating variable. or common field condition. and probability plot for each level of the accelerating variable(s) based on the fitted model. By default. Relationship: Choose a linear (no transformation). By default. For example. Failure times that occur within a certain interval of time are interval-censored. asking Minitab to extrapolate to the design value. you might want to examine how a predictor affects the lifetime of a person.

Arrhenius. lognormal. The first predictor column contains various levels of an accelerating variable. C1 Response 29 31 37 37 41 C2 Censor F F F C F C3 Count 1 19 1 1 19 C4 Factor 1 1 1 2 2 C5 Covar 12 12 12 12 12 Text categories (factor levels) are processed in alphabetical order by default. exponential. Minitab automatically excludes all observations with missing values from all calculations. although you may have more than three. Worksheet Structure for Regression with Life Data The basic worksheet structure for regression with life data is three columns. factor. For example. and identical predictor values. inverse temperature. . The second predictor column can contain either various levels of a second accelerating variable or various levels of a factor. you can define your own order − see Ordering Text Categories. or loge (power) transformation for the accelerating variable. smallest extreme value. For more information. In Variables/Start variables. as described in Failure times. censoring indicator. Assumed distribution: Choose one of eight common lifetime distributions: Weibull (default). enter the columns of failure times. see How to specify the model terms. If you wish. Factor: Enter the column containing the factor levels. enter one or two predictor columns. You can enter up to ten columns (ten different samples). and loglogistic. Minitab assumes the relationship is linear.Reliability and Survivial Analysis Second Variable: Accelerating: Enter the column containing the predictor values for the second accelerating variable. The way you set up the worksheet depends on the type of censoring you have. 142 Copyright © 2003–2005 Minitab Inc. These predictor variables may be treated as factors or covariates in the model. To perform accelerated life testing with uncensored/right censored data 1 2 Choose Stat > Reliability/Survival > Accelerated Life Testing. Structure each column so that it contains individual observations (one row = one observation). or unique observations with a corresponding column of frequencies. Here is the same worksheet structured both ways: Raw Data: one row for each observation Frequency Data: one row for each combination of response. and covariate. logistic. − For Regression with Life Data. By default. normal. if needed) predictor variables − For Accelerated Life Testing. an accelerating variable may be stresses or catalysts whose levels exceed normal operating conditions. enter one or more predictor columns. The three columns in the worksheet: • • • the response variable (failure times) censoring indicators (for the failure times. Relationship: Choose a linear (no transformation). All rights reserved. Include interaction term between variables: Check to include an interaction term between the accelerating variable and the second variable. Frequency columns are useful when you have large numbers of data with common failure and censoring times.

an inverse power relationship results. The first censoring column is paired with the first data column. columns.16 The inverse temperature transformation is a simple relationship that assumes that failure time is inversely proportional to Kelvin temperature. Arrhenius transformation = 11604. The loge (power) relationship is most often used in combination with a loge-based failure time distribution. then click OK. Note 6 7 Click Censor. All rights reserved. ? If your second predictor is a factor. and ball bearings. ? If your second predictor is a factor. semiconductor devices. the rate of a simple chemical reaction depends on the temperature. Based on the Arrhenius Rate Law. solid state devices.83 ° C + 273. enter the column of predictors.16 Copyright © 2003–2005 Minitab Inc. To perform accelerated life testing with uncensored/arbitrarily censored data 1 2 3 4 5 6 7 Choose Stat > Reliability/Survival > Accelerated Life Testing. Any change in failure time or loge failure time is directly proportional to the change in the accelerating variable. enter the columns of end times. 8 Transforming the accelerating variable If you assume a linear relationship then no transformation is needed. then click OK. enter the columns of start times. To use some other value. you can skip steps 6 & 7. enter the column containing the factor levels in Factor. and so on. Choose Responses are uncens/arbitrarily censored data. If you have a second predictor. In Variables/Start variables. enter the column containing the accelerating levels in Accelerating. If you have a second predictor. Minitab uses the lowest value in the censoring column to indicate a censored observation. enter them in Freq. In Accelerating var. Common applications of the loge transformations include electrical insulations. Inverse temperature transformation = 1 ° C + 273. The inverse and Arrhenius transformations have similar results. When it is used in combination with a loge-based failure time distribution. 8 If you like. If you have frequency columns. enter the column containing the factor levels in Factor. If you like. do one of the following ? If your second predictor is a second accelerating variable. and plastics. In Accelerating var. columns. Common applications of the Arrhenius transformation include electrical insulations. metal fatigue. enter that value in Censoring value. enter them in Freq. use one or more of the dialog box options. do one of the following ? If your second predictor is a second accelerating variable. In End variables. By default. This relationship is often used to describe failures due to degradation caused by a chemical reaction. You can enter up to ten columns (ten different samples). If you have no censored values. In Use censoring columns. use one or more of the dialog box options. the second censoring column is paired with the second data column.Accelerated Life Testing 3 4 5 If you have frequency columns. 143 . enter the censoring columns. enter the column containing the accelerating levels in Accelerating. enter the column of predictor values. You can enter up to ten columns (ten different samples). but the coefficients have different interpretations. A loge (power) relationship is used to model the life of products running under constant stress.

enter 10 50 90 (the 10th. Confidence limits: Check to store the confidence limits for the percentiles. Accelerated Life Testing . By default. Store percentiles Percentiles: Check to store the percentiles. you can ask Minitab to extrapolate information gained from the accelerated situation to the design value. Store probabilities Probabilities: Check to store the survival probabilities or cumulative failure probabilities. Because of the potentially large amount of output. Minitab store the results in the worksheet rather than printing them in the Session window. or cumulative failure probabilities for the accelerating variable(s) levels used in the study: In the Estimate subdialog box. choose. Estimate cumulative failure probabilities: Choose to estimate the cumulative failure probabilities. The first censoring column is paired with the first data column. 50th. and so on. Text values must be contained in double quotes. Estimate probabilities for times: Enter the times for which you want to estimate survival probabilities or cumulative failure probabilities. More Sometimes you may want to estimate percentiles. 144 Copyright © 2003–2005 Minitab Inc. Dialog box items Use censoring columns: If you have right censored data. Confidence intervals: Choose to use two-sided confidence intervals (the default) or just an upper or lower confidence interval. Dialog box items Percentile and Probability Estimation Enter new predictor values: Enter one or two new values or one or two columns of new values.Censor Stat > Reliability/Survival > Accelerated Life Testing > Censor Allows you to specify the censoring columns. Minitab then estimates how long it takes for 10% of the units to fail. The default is 95%. All rights reserved. The first value or column corresponds to the first variable. and 90% of the units to fail. the second censoring column is paired with the second data column. or common field condition. Minitab uses the lowest value in the censoring column to indicate a censored observation. Use predictor values in data (storage only). Accelerated Life Testing . Standard error: Check to store the standard error of the percentiles. enter that value in Censoring value.Reliability and Survivial Analysis Percentiles and survival probabilities When doing accelerated life testing. middle. Estimate survival probabilities: Choose to estimate the survival probabilities. Minitab estimates the 50th percentile. survival probabilities. enter the censoring columns. or common running condition. But most likely. Confidence level: Enter the confidence level for all of the confidence intervals. Estimate percentiles for percents: Enter the percents for which you want to estimate percentiles. Censoring value: By default. 50% of the units to fail. percentiles for given percents and survival probabilities for times. If you want to look at the beginning. the information you ultimately want is. you subject units to levels of an accelerating variable(s) far exceeding normal field conditions to accelerate the failure process. 90th percentiles). and end of the product's lifetime for a given predictor value.Estimate Stat > Reliability/Survival > Accelerated Life Testing > Estimate Allows you to estimate percentiles and survival probabilities for predictor values. Use predictor values in data (storage only): Choose to use the predictor values from the data to estimate percentiles and/or survival probabilities. How do the units behave under normal field conditions? In the Estimate subdialog box. Confidence limits: Check to store the confidence limits for the survival probabilities or cumulative failure probabilities. and the second value or column corresponds to the second variable. Often you will enter the design value. To use some other value. . for the units.

Probability plot for standardized residuals: Check to display a probability plot for standardized residuals. Diagnostic plots Probability plot for each accelerating level based on individual fits: Check to display a probability plot for each accelerating level based on individual fits. By default.Accelerated Life Testing To estimate percentiles and survival probabilities 1 2 3 In the Accelerated Life Testing dialog box. enter 10 50 90 (the 10th. You can enter one design value for each accelerating variable. 50th. Do any of the following. type the percents in the box. Exponential probability plot for Cox-Snell residuals: Check to display an exponential probability plot for Cox-Snell residuals. enter the percents in Estimate percentiles for percents. enter the times in Estimate probabilities for times. Probability plot for each accelerating level based on fitted model: Check to display a probability plot for each level of the accelerating variable based on the fitted model. or common running condition. By default. Minitab plots the 10th. Note To change the method for calculating probability plot points. • To plot percentiles for the percents you specify. Often you will enter the design value.Graphs Stat > Reliability/Survival > Accelerated Life Testing > Graphs You can draw relation plots and probability plots. enter the percents or a column of percents in Plot percentiles for percents. For example. To modify the relation plot 1 2 In the Accelerated Life Testing dialog box. Relation plot: Check to display the relationship between accelerating variable(s) and failure time using a relation plot. Accelerated Life Testing . enter one new value or column of new values. Dialog box items Plots based on fitted model Design value to include on plots: Enter a design value to include on the plots based on the fitted model (relation plot and probability plot for each accelerating level). and 90th percentiles. Minitab estimates the probability. and end of the product's lifetime for a given predictor value. 50th. Minitab plots the 10th. Display no confidence intervals: Choose to suppress display of all the confidence intervals. and 90th percentiles. enter the value in Design value to include on plot. • To estimate survival probabilities. Display confidence intervals on diagnostic plots: Check to display confidence intervals on the probability plots. 50% of the units to fail. Display confidence intervals for middle percentile: Choose to display confidence intervals for the middle percentile only. Minitab then estimates how long it takes for 10% of the units to fail. Do any of the following: • To include the design value(s) on the plot. for each predictor value. see Tools > Options > Individual Graphs > Probability Plots. Plot percentiles for percents: To plot percentiles for the percents you specify. Copyright © 2003–2005 Minitab Inc. when you enter 70 (units in hours). click Graphs. If you want to look at the beginning. For example. Display confidence intervals for design value: Choose to display confidence intervals for design value. By default. and 90% of the units to fail. for the units. 145 . click Estimate. then click OK: • To estimate percentiles. Display confidence intervals for all levels: Choose to display confidence intervals for all levels. that the unit will survive past 70 hours. Minitab estimates the 50th percentile. 50th. enter 30. Display confidence intervals for all percentiles: Choose to display confidence intervals for all percentiles. All rights reserved. Display no confidence intervals: Choose to suppress display of all the confidence intervals. In Enter new predictor values. 90th percentiles). Display failure times on plot: Check to display points for failure times (exact failure time or midpoint of interval for interval censored observation) on the plot. middle. to plot the 30th percentile (how long it takes 30% of the units to fail).

For an illustration. see Probability plots. The probability plot based on the fitted model includes fitted lines that are based on the chosen distribution and transformation. . If you like. A count of failures and right-censored data appears when your data are exact failures/right-censored. You can draw probability plots for the standardized and Cox-Snell residuals. By default. change the confidence level for the intervals (default = 95%): Click Estimate. For more information on creating and interpreting probability plots see Probability plots. one relation plot is displayed. This plot assumes that the observations for each level of the accelerating variable share a common shape (Weibull and exponential distributions) or scale (other distributions). for all levels of the accelerating variable(s). 50th. see Example of accelerated life testing. Probability plot for each accelerating level based on fitted model The probability plot displays the percents for each level of the accelerating variable(s) based on the fitted model (line) and a nonparametric model (points). You can enter design value(s) to include on the plot(s). and 90th percentiles. then consider a different transformation or distribution. When you have one accelerating variable and one factor. All rights reserved. which is a goodness-of-fit measure. By default. The accelerating variable is on the x-axis and is held at each level of the factor.Reliability and Survivial Analysis • Choose one: − − − Display confidence intervals for middle percentile Display confidence intervals for all percentiles Display no confidence intervals • 3 4 To include failure times (exact failure time or midpoint of interval for interval censored observation) on the plot. or no confidence intervals. You can use the Anderson-darling statistic to compare the fit of competing models. check Display failure times on plot. The 50th percentile is a good estimate for the time a part will last for the given conditions. You can optionally specify up to 10 percentiles to plot and display the failure times (exact failure time or midpoint of interval for interval censored observation) on the plot. In Confidence level. enter a value. the better the fit. If the points do not fit the lines adequately. 146 Copyright © 2003–2005 Minitab Inc. When you have two accelerating variable. transformation. You can choose to display confidence intervals for the design value(s). then the fitted lines should be approximately parallel. The first plot contains the first accelerating variable on the x-axis and is held at each level of the second accelerating variable. In general. Probability plots The Accelerated Life Testing command draws several probability plots to help you assess the fit of the chosen distribution. and assumption of equal shape (Weibull or exponential) or scale (other distributions) are appropriate. For a discussion of probability plots. The probability plot based on the individual fits includes fitted lines that are calculated by individually fitting the distribution to each level of the accelerating variable. two relation plots are displayed. You can also choose to draw probability plots for each level of the accelerating variable based on individual fits or on the fitted model. then click OK. A smaller Anderson-Darling statistic indicates that the distribution provides a better fit. lines are drawn at the 10th. The relation plot(s) displayed depend on the number of predictors in your model: • • • When you have one accelerating variable. These parameters are based on the fitted model. You can use these plots to assess whether a particular distribution fits your data. You can use these plots to assess whether the distribution. Relation plot The relation plot displays failure time versus an accelerating variable. the probability plot includes the shape and scale parameters (Weibull and exponential distributions) or the location and scale parameters (other distributions). The points should fit the line adequately if the chosen distribution is appropriate. Click OK. The probability plot also includes the Anderson-Darling statistic. the closer the points fall to the fitted line. one relation plot is displayed. Minitab provides one goodness-of-fit measure: the Anderson-Darling statistic. If the distributions have equal shape (Weibull or exponential) or scale (other distributions) parameters.

the regression table.Options Stat > Reliability/Survival > Accelerated Life Testing > Options You can estimate the model parameters from the data or enter historical estimates − see Estimating the Model Parameters for more information. or a separate column for each response variable. Maximum number of iterations: Enter a positive integer to specify the maximum number of iterations for the NewtonRaphson algorithm. The column(s) should contain one value for each coefficient in the regression table.Results Stat > Reliability/Survival > Accelerated Life Testing > Results You can control the display of Session window output. censoring information. Accelerated Life Testing . in the order that the coefficients appear in the regression table. Copyright © 2003–2005 Minitab Inc. Variance/covariance matrix: Check to store the variance/covariance matrix for the estimated coefficients. table of percentiles.Accelerated Life Testing Accelerated Life Testing . Set shape (Weibull) or scale (other distributions) at: To estimate other model coefficients while holding the shape or scale parameter fixed. plus a table of percentiles and/or survival probabilities (if you requested them in Accelerated Life Testing . or a number of values equal to the number of response variables. survival probabilities. Ordinary Standardized Cox-Snell Information on Estimated Equation Estimated coefficients: Check to store the estimated coefficients. and goodness-of-fit: Choose to display the response and censoring information. and goodness-of-fit measures. Use starting estimates: If you have starting estimates. log-likelihood. Enter one column to be used for all of the response variables. Response information. Dialog box items Control the Display of Results Display nothing: Choose to suppress all printed output. but do all requested storage and display graphs. Standard error of estimates: Check to store the standard error of the estimated coefficients. In addition. Log-likelihood for last iteration: Check to store the log-likelihood for the last iteration. All rights reserved. list of factor level terms.Estimate). The column(s) should contain one value for each coefficient in the regression table. enter one column to be used for all of the response variables. in the order that the coefficients appear in the regression table. the log-likelihood. Dialog box items Residuals: Check any of the residual types below to store them in the worksheet. enter one value to be used as the shape or scale parameter for all of the response variables. and tests for terms with more than 1 degree of freedom. list of factor level values and tests for terms with more than 1 degree of freedom: Choose to display the output described above. Use historical estimates: Choose to enter your own estimates for the model parameters. Accelerated Life Testing . or a separate column for each response variable.Storage Stat > Reliability/Survival > Accelerated Life Testing > Storage You can store three types of residuals and information on the estimated equation. regression table. Show log-likelihood for each iteration of algorithm: Check to display the log-likelihood at each iteration of the parameter estimation process. Confidence limits for coefficients: Check to store the confidence limits for the coefficients. Dialog box items Estimate model parameters: Choose to estimate the model parameters from the data. 147 .

82462 Z -15. In Variables/Start variables.000 Log-Likelihood = -564. enter Design.256969 95. It is known that an Arrhenius relationship exists between temperature and failure time. From Relationship. then click OK in each dialog box. 1 2 3 4 5 6 7 Open the worksheet INSULATE. To see how well the model fits.0% Normal CI Lower Upper -17. In Enter new predictor values. enter Censor. Session window output Accelerated Life Testing: FailureT versus Temp Response Variable: FailureT Censoring Information Uncensored value Right censored value Count 66 14 Censoring value: Censor = C Estimation Method: Maximum Likelihood Distribution: Weibull Arrhenius Relationship with accelerating variable(s): Regression Table Standard Error 0. All rights reserved. enter Temp. Click Censor.762042 0. choose Arrhenius. The motors normally run between 80 and 100° C.6 148 Copyright © 2003–2005 Minitab Inc. and 170° C − to speed up the deterioration.0350418 0. In Accelerating variable. In Use censoring columns. Click Estimate. With failure time information at these temperatures. In Design value to include on plot. enter 80.51 95.0% Normal CI Lower Upper 113918 223557 29543.MTW. you decide to use accelerated life testing. First you gather failure times for the insulation at abnormally high temperatures − 110. Choose Stat > Reliability/Survival > Accelerated Life Testing.1203 -13.4 46209. Click OK.000 0. you can then extrapolate to 80 and 100° C.71 P 0. Click Graphs.2546 0.899403 2. 150.693 Table of Percentiles Standard Error 27446. .40 23. then click OK.830722 2.9 4216.9 Percent 50 50 Temp 80 100 Percentile 159584 36948. you will draw a probability plot based on the standardized residuals.986180 0.36332 3. 130.Reliability and Survivial Analysis Example of Accelerated Life Testing Suppose you want to investigate the deterioration of an insulation used for electric motors. To save time and money. enter FailureT.37596 Predictor Intercept Temp Shape Coef -15.1874 0.

149 .Accelerated Life Testing Graph window output: Copyright © 2003–2005 Minitab Inc. All rights reserved.

the Scale parameter. at 100° C. and assumption of equal shape (Weibull) at each level of the accelerating variable are appropriate. Regression table The regression table displays: • the estimated coefficients for the regression model and their − standard errors.8246 εp where εp = the pth percentile of the standard extreme value distribution ArrTemp = 11604. All rights reserved. and its − standard error.21 years. or 4. you get the coefficients for the regression model. the Shape parameter (Weibull or exponential) or Scale parameter (other distributions). the insulation lasts about 159.83 Temp + 273. − the log-likelihood. . the 10th.584.83072 (ArrTemp) + 1/2. this model describes the relationship between temperature and failure time for the insulation: Loge (failure time) = -15.57 hours. Output The default output consists of the regression table and relation plot. • • • 150 Copyright © 2003–2005 Minitab Inc. 50th.20 years. The probability plot based on the fitted model can help you determine whether the distribution. you can look at the distribution of failure times for each temperature − in this case. − 95% confidence interval. In this case.948. a measure of the overall variability.Reliability and Survivial Analysis Interpreting the results From the Regression Table. The 50th percentile is a good estimate of how long the insulation will last in the field. the points fit the lines adequately. is it a significant predictor? − 95% confidence interval. thereby verifying that the assumptions of the model are appropriate for the accelerating variable levels. a measure of the overall variability. Anderson-Darling goodness-of-fit statistics for each level of the accelerating variable based on the fitted model. in other words. transformation. or 18.16 The Table of Percentiles displays the 50th percentiles for the temperatures that you entered. the insulation lasts about 36. − Z-values and p-values. With the relation plot. At 80° C.1874 + 0. − 95% confidence interval.5 hours. and its − standard error. For a Weibull distribution. The Z-test tests that the coefficient is significantly different than 0. and 90th percentiles.

• Both regression with life data commands differ from other regression commands in Minitab in that they use different distributions and accept censored data. you must have a good model of the relationship between failure time and the accelerating variable(s). you may only know that a product failed before a certain time. Some of these variables may be factors. Similarly. Regression with Life Data differs from Minitab's regression commands in that it accepts censored data and uses different distributions. For example. logistic. part. for example why some products fail quickly and some survive for a long time. The first predictor is an accelerating variable. Model: Enter the model terms − see How to specify the model terms. Regression with Life Data Stat > Reliability/Survival > Regression with Life Data Use Regression with Life Data to see whether one or more predictors affect the failure time of a product. This model will help you understand how different factors and covariates affect the lifetime of your part or product. Responses are uncens/ arbitrarily censored data: Choose if your data is uncensored or arbitrarily censored. exponential. The model can include factors. • Accelerated Life Testing performs a simple regression with one or two predictors that is used to model failure times for highly reliable products. columns (optional): Enter a column for each variable containing the frequency data.Regression with Life Data Regression with Life Data Regression with Life Data Overview Use Minitab's regression with life data commands to investigate the relationship between failure time and one or more predictors. logistic. Life data is often incomplete or censored in some way. its levels exceed those normally found in the field. 151 . If any of those predictors are factors. interactions. normal. you must enter the following information: • • the response variable (failure times). Failure times that occur within a certain interval of time are interval-censored. The goal is to come up with a model that predicts failure time. smallest extreme value. and nested terms. covariates. Assumed distribution: Choose one of eight common lifetime distributions: Weibull (default). covariates. See How to specify the model terms. normal. In order to do this. Any products that have not failed before the study ends are right-censored. This model uses explanatory variables to explain changes in the response variable. smallest extreme value. Censored observations are those for which an exact failure time is unknown. lognormal. and loglogistic. lognormal basee. various interactions between predictors and nested terms. Dialog box items Responses are uncens/right censored data: Choose if your data is uncensored or right censored. which consist of any number of predictor variables and when appropriate. Variables/Start variables: Enter up to 10 columns (10 different samples) containing the start times. enter them again in Factors. Data . End variables: If you have uncensored or arbitrarily censored data. and nested terms. The goal is to come up with a model that predicts failure time. Freq. To do regression with life data. interactions. you might want to examine how a predictor affects the lifetime of a person. All rights reserved. You can choose to model your data on one of the following eight distributions: Weibull. Copyright © 2003–2005 Minitab Inc. Factors (optional): Enter any variables in the model that are factors.Regression with Life Data Enter three types of columns in the worksheet: • the response variable (failure times) − see Failure times. The data obtained under the high stress conditions can then be used to extrapolate back to normal use conditions. model terms. Based on these predictions you can estimate the reliability of the system. which is left-censored. meaning that the part failed sometime after the present time. The second predictor can be either a second accelerating variable or a factor. Suppose you are testing how long a product lasts and you plan to end the study after a certain amount of time. exponential. The model can include factors. Regression with Life Data performs a regression with one or more predictors. product. or organism. enter up to 10 columns (10 different samples) of end times. Minitab uses a modified Newton-Raphson algorithm to calculate maximum likelihood estimates of the model parameters. and loglogistic.

Minitab estimates the coefficient associated with the covariate to describe its effect on the response variable. Failure time ∗ (missing value symbol) Left censored Interval censored time before which the failure occurred time at the end of interval during which the failure occurred This example uses a frequency column as well. which may be factors (categorical variables) or covariates (continuous variables). For covariates... see Worksheet Structure for Regression with Life Data. Unless you specify a predictor as a factor. to compare the effect of different levels on the response variable. 152 Copyright © 2003–2005 Minitab Inc. How you run the analysis depend on whether your data are uncensored/right censored or uncensored/arbitrarily censored: Uncensored/arbitrarily censored data If you have any combination of exact failure times. Minitab by default designates the lowest numeric or text value as the reference level. or as frequency data. Factor columns can be numeric or text. Minitab estimates the coefficients for k . Depending on the type of censoring you have. You can also structure the worksheet as raw data.. you will set up your worksheet in column or table form. or nested within factors. left. Minitab automatically excludes all observations with missing values from all calculations. Exact failure time Right censored Enter in the Start column. For factors. Factors may be crossed or nested. see Factor variables and reference levels.. The model can include up to 9 factors and 50 covariates. 2 units are exact failures at 30000 hours. Covariates may be crossed with each other or with factors. All rights reserved. if needed predictor variables. . To change the reference level. Start End Frequency 20 units are left censored at 10000 hours. the predictor is assumed to be a covariate. enter your data using a Start column and End column: For this observation. or nested terms. In the model.and interval-censored data.. See How to specify the model terms. 190 units are right censored at 90000 hours. interactions.. right-.1 design variables (where k is the number of levels). covariates.Reliability and Survivial Analysis • • censoring indicators for the response variables. Failure time time after which the failure occurred ∗ (missing value symbol) time at start of interval during which the failure occurred Enter in the End column. For details. 50 units are interval censored between 50000 and 60000 hours. You can enter up to ten samples per analysis. terms may be created from these predictor variables and treated as factors.

Minitab assumes the lower of the two values indicates censoring. You only know that the failure occurred before a particular time. etc. and the higher of the two values indicates an exact failure. text. Interval censored How you set up your worksheet depends. Censor F F F F F C F F . Left censored The fan failed sometime before 500 days. etc. You only know that the failure occurred after a particular time. or date/time values. Individual failure times are the same type of data used for Distribution Analysis. . When your data have a varied censoring scheme. a C designates a unit that was removed from the test. Right censored The fan had not yet failed at 500 days. and was thus censored. 60 53 40 51 99 35 53 . You only know that the failure occurred between two particular times. in part. The data column and associated censoring column must be the same length. or under varying conditions of any combination of accelerating variables. The Censor column contains the corresponding censoring indicators: an F designates an actual failure time. 153 . The fan failed sometime between 475 and 500 days. . although pairs of data/censor columns (each pair corresponds to a sample) can have different lengths. you might collect failure times for units running at a given temperature. If you do not specify which value indicates censoring in the Censor subdialog box. The table below describes the types of observations you can have: Type of observation Exact failure time Description You know exactly when the failure occurred. . Life data is often censored or incomplete in some way. Failure times The response data you gather for the regression with life data commands are the individual failure times. on the type of censoring you have: • • When your data consist of exact failures and right-censored observations. . see Uncensored/right censored data.Regression with Life Data Uncensored/right censored data Enter two columns for each sample − one column of failure (or censoring) times and a corresponding column of censoring indicators. For example. see Uncensored/arbitrarily censored data. All rights reserved. Censoring indicators can be numbers. Example The fan failed at exactly 500 days. Time 53 The Time column contains failure times. You might also collect samples under different temperatures. Copyright © 2003–2005 Minitab Inc. Suppose you are monitoring air conditioner fans to find out the percentage of fans that fail within a three-year warranty period.

When you let Minitab estimate the parameters from the data. 154 Copyright © 2003–2005 Minitab Inc. In End variables. enter a column with entries that correspond to the model terms in the order you entered them in the Model box. all results − such as the percentiles − are based on these parameters. If you have no censored data. change the maximum number of iterations for reaching convergence (the default is 20). so you may want to specify what you think are good starting values for parameter estimates. enter the censoring columns. Factor variables and reference levels You can enter numeric. The first censoring column is paired with the first data column. . then click OK. enter that value in Censoring value. you can optionally: • • enter starting values for the algorithm. Note 5 6 Click Censor. Minitab obtains maximum likelihood estimates through an iterative process. If any of those predictors are factors. no estimation is done. text. enter them again in Factors. enter them in Freq. In Variables/Start variables. find out the order of entries for the starting estimates column by looking at the regression table in the output. use any dialog box options. To use some other value. Choose Responses are uncens/arbitrarily censored data. enter the model terms − see How to specify the model terms. In Variables/Start variables. Estimating the model parameters Minitab uses a modified Newton-Raphson algorithm to estimate the model parameters.Reliability and Survivial Analysis To perform regression with uncensored/right censored data 1 2 3 4 Choose Stat > Reliability/Survival > Regression with Life Data. If you have frequency columns. enter up to ten columns of start times (ten different samples). • Why enter starting values for the algorithm? The maximum likelihood solution may not converge if the starting estimates are not in the neighborhood of the true solution. In Model. and so on. estimate other model coefficients while holding the shape parameter (Weibull) or the scale parameter (other distributions) fixed at a specific value. columns. enter them in Freq. columns. With complicated models. you can enter your own parameters. In this case. If any of those predictors are factors. the second censoring column is paired with the second data column. then click OK. If you like. enter the model terms − see How to specify the model terms. By default. Minitab uses the lowest value in the censoring column to indicate a censored value. In Use censoring columns. 7 If you like. enter them again in Factors. enter up to ten columns of failure times (ten different samples). enter up to ten columns of end times (ten different samples). meaning that the estimated coefficients are interpreted relative to this level. the command terminates. If the maximum number of iterations is reached before convergence. If you have frequency columns. If you like. or date/time factor levels. you can skip steps 5 & 6. use any dialog box options. To perform regression with uncensored/arbitrarily censored data 1 2 3 4 5 6 7 Choose Stat > Reliability/Survival > Regression with Life Data. In Model. Minitab assigns one factor level to be the reference level. In all cases. All rights reserved.

By default. Most often. Estimate percentiles for percents: Enter the percents for which you want to estimate percentiles. Minitab uses the lowest value in the censoring column to indicate a censored observation. If you want to look at the beginning. the second censoring column is paired with the second data column. enter the censoring columns.Regression with Life Data Regression with Life Data creates a set of design variables for each factor in the model. If you like. you can request a multiple degrees of freedom test. Minitab estimates the 50th percentile. or text value as the reference factor level. Copyright © 2003–2005 Minitab Inc. middle. Estimate survival probabilities: Choose to estimate survival probabilities. Use predictor values in data (storage only): Choose to use the predictor values from the data to estimate percentiles and/or survival probabilities. This procedure tests whether or not the term is significant. enter 10 50 90 (the 10th. you would enter the design value. 50% of the units to fail. and so on. 50th. and 90% of the units to fail. Values (or columns) should be listed in the order that the corresponding variables appear in the model.1 design variables and the reference level will be coded as 0. Confidence limits: Check to store the confidence limits for the survival probabilities or cumulative failure probabilities. Standard error: Check to store the standard error of the percentiles. 90th percentiles). Store probabilities Probabilities: Check to store the survival probabilities or cumulative failure probabilities. To use some other value. Here are two examples of the default coding scheme: Factor A with 4 levels (1 2 3 4) reference level is 1 A1 1 2 3 4 0 1 0 0 A2 0 0 1 0 A3 0 0 0 1 reference level is High Factor B with 3 levels (High Low Medium) B1 High Low Medium 0 1 0 B2 0 0 1 By default. date/time. there will be k . you can change this reference value in the Options subdialog box. Estimate cumulative failure probabilities: Choose to estimate cumulative failure probabilities. Estimate probabilities for times: Enter the times for which you want to estimate survival probabilities or cumulative failure probabilities. or one or more columns. In other words: Is at least one of the coefficients associated with this term significantly different than zero? Regression with Life Data . Minitab then estimates how long it takes for 10% of the units to fail. Dialog box items Use censoring columns: If you have right censored data. Censoring value: By default. The first censoring column is paired with the first data column. enter that value in Censoring value. and end of the product's lifetime for a given predictor value. Minitab designates the lowest numeric. You can enter the values separated by spaces. All rights reserved. Confidence limits: Check to store the confidence limits for the percentiles.Estimate Stat > Reliability/Survival > Regression with Life Data > Estimate Percentile and Probability Estimation Enter new predictor values: Choose to specify values for predictors to estimate percentiles and/or survival probabilities. 155 . If there are k levels. Text values must be contained in double quotes. Regression with Life Data . Multiple degrees of freedom test When you have a term with more than one degree of freedom. Store percentiles Percentiles: Check to store the percentiles.Censor Stat > Reliability/Survival > Regression with Life Data > Censor Allows you to specify the censoring columns.

. If you want to look at the beginning. 3 Regression with Life Data . that the unit will survive past 70 hours. For an illustration. the closer the points fall to the fitted line. You can use these plots to assess whether a particular distribution fits your data. 1 2 In the Regression with Life Data dialog box. middle. 50% of the units to fail. click Estimate. choose Use predictor values in data (storage only). Confidence intervals: Choose to use two-sided confidence intervals (the default) or just an upper or lower confidence interval. and end of the product's lifetime for a given predictor value. or the values in your data. a smaller Anderson-Darling statistic indicates that the distribution provides a better fit. The default is 95%. the better the fit. check Probability plot for standardized residuals • To plot the Cox-Snell residuals. click Graphs. see Example of regression with life data. It measures the distances from the plot points to the fitted line. Do one of the following: • To enter new predictor values: In Enter new predictor values. Minitab then estimates how long it takes for 10% of the units to fail.Graphs Stat > Reliability/Survival > Regression with Life Data > Graphs You can draw probability plots for the standardized and Cox-Snell residuals. Probability plots for regression with life data The Regression with Life Data command draws probability plots for the standardized and Cox-Snell residuals. • To use the predictor values in the data. Do any of the following. To draw a probability plot of the residuals 1 2 In the Accelerated Life Testing or Regression with Life Data dialog box. then use the probability plot included with the Parametric Distribution Analysis commands. 50th. Dialog box items Probability plot for standardized residuals: Check to display a probability plot for standardized residuals. For example. The predictor values must be in the same order as the main effects in the model. Note 156 Copyright © 2003–2005 Minitab Inc.Reliability and Survivial Analysis Confidence level: Enter the confidence level for all of the confidence intervals. enter the percents or a column of percents in Estimate percentiles for percents. To estimate percentiles and survival probabilities You can estimate percentiles and survival probabilities for new predictor values. Because of the potentially large amount of output. Note To change the method for calculating probability plot points. All rights reserved. enter 10 50 90 (the 10th. then click OK: • To plot the standardized residuals. then click OK: • To estimate percentiles. Display confidence intervals on probability plots: Check to display confidence intervals on the probability plots. Exponential probability plot for Cox-Snell residuals: Check to display an exponential probability plot for Cox-Snell residuals. In general. when you enter 70 (units in hours in this example). and 90% of the units to fail. enter the times or a column of times in Estimate probabilities for times. By default. see Tools > Options > Individual Graphs > Probability Plots. Do any of the following. Minitab provides one goodness-of-fit measure: the Anderson-Darling statistic. store the residuals in the Storage subdialog box. • To estimate survival probabilities. 90th percentiles). check Exponential probability plot for Cox-Snell residuals To draw a probability plot with more options. Minitab stores the results in the worksheet rather then printing them in the Session window. for each predictor value. Minitab estimates the probability. therefore. enter a set of predictor values (or columns containing sets of predictor values) for which you want to estimate percentiles or survival probabilities. Minitab estimates the 50th percentile. The Anderson-Darling statistic is useful in comparing the fit of different distributions.

To control estimation of the parameters 1 2 In the Regression with Life Data dialog box. but do all requested storage and display graphs. Response information. Enter one column to be used for all of the response variables. Click OK. The column(s) should contain one value for each coefficient in the regression table. Set shape (Weibull) or scale (other distributions) at: To estimate other model coefficients while holding the shape or scale parameter fixed. or a number of columns equal to the number of response variables. − − − To enter starting estimates for the parameters: In Use starting estimates. Maximum number of iterations: Enter a positive integer to specify the maximum number of iterations for the NewtonRaphson algorithm. To specify the Maximum number of iterations. choose Use historical estimates and enter one column to be used for all of the response variables. or a number of columns equal to the number of response variables. enter a factor column followed by a value specifying the reference level. the regression table. enter one column to be used for all of the response variables. enter a factor column followed by a value specifying the reference level. You can also change the reference level for a factor. For date/time values.Regression with Life Data Regression with Life Data . store the value as a constant and then enter the constant. click Options. censoring information. for each factor you want to set the reference level for. and goodness-of-fit measures. or a separate column for each response variable. Click OK. or a number of values equal to the number of response variables. and goodness-of-fit: Choose to display the response and censoring information.Results Stat > Reliability/Survival > Regression with Life Data > Results You can control the display of Session window output. 157 . Do one of the following: • To estimate the model parameters from the data (the default). or a separate column for each response variable. Use starting estimates: If you have starting estimates. regression table. in the order that the coefficients appear in the regression table. For text values. log-likelihood. In Reference factor level. For text values. To change the reference factor level 1 2 In the Regression with Life Data dialog box. • 3 To enter your own estimates for the model parameters. To estimate other model coefficients while holding the shape parameter (Weibull) or the scale parameter (other distributions) fixed: In Set shape (Weibull) or scale (other distributions) at. choose Estimate model parameters. For date/time values. Reference factor level (enter factor followed by level): For each factor you want to set the reference level for. the log-likelihood. enter one column to be used for all of the response variables. enter one value to be used for all of the response variables. the value must be in double quotes. the value must be in double quotes. store the value as a constant and then enter the constant. The column(s) should contain one value for each coefficient in the regression table. or a number of values equal to the number of response variables. enter a positive integer. click Options. 3 Regression with Life Data .Options Stat > Reliability/Survival > Regression with Life Data > Options You can estimate the model parameters from the data or enter historical estimates − see Estimating the Model Parameters for more information. Use historical estimates: Choose to enter your own estimates for the model parameters. enter one value to be used as the shape or scale parameter for all of the response variables. Copyright © 2003–2005 Minitab Inc. Dialog box items Estimate model parameters: Choose to estimate the model parameters from the data. All rights reserved. Dialog box items Control the Display of Results Display nothing: Choose to suppress all printed output. in the order that the coefficients appear in the regression table.

list of factor level values. In Model. table of percentiles and/or survival probabilities: Choose to display the output described above. click Results.MTW. Click Censor. tests for terms with more than one degree of freedom: Choose to display all of the output described above. list of factor level values. you will draw a probability plot based on the standardized residuals. . Confidence limits for coefficients: Check to store the confidence limits for the coefficients. Show log-likelihood for each iteration of algorithm: Check to display the log-likelihood at each iteration of the parameter estimation process. 150. enter FailureT. To perform multiple degrees of freedom tests 1 2 In the Regression with Life Data dialog box.Storage Stat > Reliability/Survival > Regression with Life Data > Storage You can store three types of residuals and information on the estimated equation. Dialog box items Residuals Check any of the residual types below to store them in the worksheet. Click Graphs. Because the motors generally run at between 80 and 100° C. then click OK. then click OK in each dialog box. Check Probability plot for standardized residuals. You gather failure times at plant 1 and plant 2 for the insulation at four temperatures − 110. then click OK. Choose In addition. Click Estimate. enter Plant. plus a list of the factor level values. Example of Regression with Life Data Suppose you want to investigate the deterioration of an insulation used for electric motors. In Factors (optional). and a multiple degrees of freedom test. Standard error of estimates: Check to store the standard error of the estimated coefficients. Choose Stat > Reliability/Survival > Regression with Life Data. 1 2 3 4 5 6 7 Open the worksheet INSULATE. and the temperature at which the motor runs. enter Censor.Reliability and Survivial Analysis In addition. All rights reserved. Variance/covariance matrix: Check to store the variance/covariance matrix for the estimated coefficients.Estimate). Ordinary Standardized Cox-Snell Information on Estimated Equation Estimated coefficients: Check to store the estimated coefficients. you want to predict the insulation's behavior at those temperatures. 158 Copyright © 2003–2005 Minitab Inc. In Variables/Start variables. It is known that an Arrhenius relationship exists between temperature and loge failure time. In Enter new predictor values. In Use censoring columns. plus a table of percentiles and/or survival probabilities (if requested in Regression with Life Data . enter ArrNewT NewPlant. and 170° C. enter ArrTemp Plant. then click OK. tests for terms with more than 1 degree of freedom. Regression with Life Data . To see how well the model fits. Log-likelihood for last iteration: Check to store the log-likelihood for the last iteration. 130. You want to know if you can predict failure times for the insulation based on the plant in which it was manufactured. In addition.

905837 -0.0% Normal CI Lower Upper 128390 258261 109690 210578 32548.8600 32.2047 -13.14 P 0.76 3913.52439 Predictor Intercept ArrTemp Plant 2 Shape Coef -15.525 Table of Percentiles Standard Error 32466.0339710 0.5 Copyright © 2003–2005 Minitab Inc.0% Normal CI Lower Upper -17.45768 -0.Regression with Life Data Session window output Regression with Life Data: FailureT versus ArrTemp.839255 -0.71 -2.2 25286.4775 0.033 Log-Likelihood = -562. 159 .346525 2.9 27781.0 43248. All rights reserved.87 95.6 5163.0845721 0.270658 95.4 52990.3411 0.000 0.180767 2.0988 31.4 34662.8600 31.000 0.6 Percent 50 50 50 50 ArrTemp 32.772673 0.94309 Z -16.0988 Plant 1 2 1 2 Percentile 182094 151981 41530.13 24.950822 0. Plant Response Variable: FailureT Censoring Information Uncensored value Right censored value Count 66 14 Censoring value: Censor = C Estimation Method: Maximum Likelihood Distribution: Weibull Linear Relationship with accelerating variable(s): Regression Table Standard Error 0.0150083 3.

respectively: Loge (failure time) = −15.9431 Loge (failure time) = −15. For motors running at 100° C. 160 Copyright © 2003–2005 Minitab Inc.9431 where = the pth percentile of the error distribution ArrTemp = 11604.52187 + 0. As you can see from the low p-values. insulation from plant 2 lasts about 151980.95 years. For the Weibull distribution. All rights reserved.51 hours or 3.83925 (ArrTemp) + 1/2. insulation from plant 2 lasts about 34662.38 hours or 4. Here. insulation from plant 1 lasts about 182093. transformation.05 level.83925 (ArrTemp) + 1/2.34 years. here is the equation that describes the relationship between temperature and failure time for the insulation for plant 1 and 2. The 50th percentile is a good estimate of how long the insulation will last in the field: • • For motors running at 80° C.Reliability and Survivial Analysis Graph window output: Interpreting the results From the Regression Table.6 hours or 20. the plot points fit the fitted line adequately.74 years.16) The Table of Percentiles displays the 50th percentiles for the combinations of temperatures and plants that you entered.8 hours or 17. and temperature is a significant predictor. insulation from plant 1 lasts about 41530. The probability plot for standardized residuals will help you determine whether the distribution. the plants are significantly different at the α = . . you get the coefficients for the regression model. therefore you can assume the model is appropriate.3411 + 0.83/(Temp + 273. and equal shape (Weibull or exponential) or scale parameter (other distributions) assumption is appropriate.77 years.

a measure of the overall variability. and its − standard error. in other words. − Z-values and p-values. All rights reserved. which displays: • the estimated coefficients for the regression model and their − standard errors. the log-likelihood. and its − standard error. 161 . • • • Copyright © 2003–2005 Minitab Inc.Regression with Life Data Default output The default output consists of the regression table. The Z-test tests that the coefficient is significantly different than 0. is it a significant predictor? − 95% confidence interval. − 95% confidence interval. a measure of the overall variability. − 95% confidence interval. the Scale parameter. the Shape parameter (Weibull or exponential) or Scale parameter (other distributions).

.

logistic. what's the probability that an insect dies? Probit Analysis Stat > Reliability/Survival > Probit Analysis Use probit analysis when you want to estimate percentiles. what shock level cracks 10% of the hulls? What concentration of a pollutant kills 50% of the fish? Or. at a given pesticide application. Probit analysis differs from accelerated life testing in that the response data is binary (success or failure). lognormal. All rights reserved. with frequency (optional): Optionally. Response: Enter the column that contains the response data. Stress (stimulus): Enter a column of stress or stimulus levels. and draw probability plots. Data . then record whether or not the hull cracked. When you enter a factor and choose a Weibull. Choices include the normal (default). Minitab calculates the model coefficients using a modified Newton-Raphson algorithm. In the engineering sciences. 163 . enter a column of frequencies. rather than an actual failure time. Number of trials: Enter the column containing the number of trials. you can also compare the potency of the stress under different conditions. a common experiment would be destructive inspecting. and cumulative failure probabilities for the distribution of a stress. where you subject organisms to various levels of a stress and record whether or not they survive. a common experiment would be the bioassay. survival probabilities. Weibull.Probit Analysis Probit Analysis Probit Analysis Overview A probit study consists of imposing a stress (or stimulus) on a number of units. You subject the materials to various magnitudes of explosions. lognormal. Probit analysis can answer these kinds of questions: For each hull material. loglogistic. then recording whether the unit failed or not. and smallest extreme value distributions.Probit Analysis Enter the following columns in the worksheet: • • • two columns containing the response variable. Number of successes: Enter the column containing the number of successes. Response in response/frequency format: Choose if your data is in response/frequency format. Factor (optional): Enter one column of factor levels. set up in success/trial or response/frequency format one column containing a stress variable (treated as a covariate in Minitab) (optional) one column containing a factor Copyright © 2003–2005 Minitab Inc. Suppose you are testing how well submarine hull materials hold up when exposed to underwater explosions. In the life sciences. Assumed distribution: Click the drop-down arrow and choose one of seven common lifetime distributions. Dialog box items Response in success/trial format: Choose if your data is in success/trial format. or loglogistic distribution.

then click OK. In Stress (stimulus). The higher value corresponds to a success. The Frequency column indicates how many times that observation occurred. For example. In Number of trials. You can take the log of the stress to get the lognormal. 1 2 Choose Stat > Reliability/Survival > Probit Analysis. 1 2 • 1 2 3 4 In Number of successes. For example. use any of the dialog box options. respectively. Do one of the following: • Choose Responses in success/trial format. Choose Responses in response/frequency format. enter one column of response values. . If you wish. To perform a probit analysis How you run the analysis depend on whether your worksheet is in "success/trial" or "response/frequency" format. enter the column in with frequency. and extreme value − allowing you to fit a broad class of binary response models. You can enter the data in either success/trial or response/frequency format. Factors Text categories (factor levels) are processed in alphabetical order by default. In Response. and Weibull distributions. you can define your own order − see Ordering Text Categories. If you like. enter one column of trials. success or failure. when Temp = 140. If you have a frequency column. when Temp = 160. This class of models (for the situation with no factor) is defined by: where = the probability of a response for the jth stress level g(yj) β0 xj = the distribution function (described below) = the constant term = the jth level of the stress variable 164 Copyright © 2003–2005 Minitab Inc. enter one column of stress or stimulus levels. enter one column of successes. there were nine successes and one failure. Probit model and distribution function Minitab provides three main distributions − normal. All rights reserved. The Trials column contains the number of observations. loglogistic.Reliability and Survivial Analysis Response variable The response data is binomial. there were seven successes and three failures. logistic. Here is the same data arranged both ways: Success/trial format Temp 80 120 140 160 Success 2 4 7 9 Trials 10 10 10 10 Response/frequency format Temp 80 80 120 120 140 140 160 160 Response 1 0 1 0 1 0 1 0 Frequency 2 8 4 6 7 3 9 1 The Success column contains the number of successes. so you have two possible outcomes. The Response column contains values which indicate whether the unit succeed or failed.

Certain distributions may be used for historical reasons or because they have a special meaning in a discipline. Here are two examples of the default coding scheme: Factor A with 4 levels (1 2 3 4) reference level is 1 A1 1 2 3 4 0 1 0 0 A2 0 0 1 0 A3 0 0 0 1 reference level is High Factor B with 3 levels (High Low Medium) B1 High Low Medium 0 1 0 B2 0 0 1 By default. you may want to consider the fact that the stress does not cause all of the deaths in the analysis. text. or to use as a starting value for the algorithm. If you like. Why enter starting values for the algorithm? The maximum likelihood solution may not converge if the starting estimates are not in the neighborhood of the true solution. the command terminates. If the natural response rate is greater than 0. For example. Minitab designates the lowest numeric. All rights reserved. This statistic is used in situations with high mortality or high failure rates. no estimation is done. you might want to know the probability that a young fish dies without being exposed to a certain pollutant. If the maximum number of iterations is reached before convergence. you can optionally: • • enter starting values for the algorithm. Natural response rate The regression table includes the natural response rate − the probability that a unit fails without being exposed to any of the stress. Goodness-of-fit statistics can be used to compare fits using different distributions. 165 . If there are k levels. meaning that the estimated coefficients are interpreted relative to this level. or text value as the reference factor level. When you let Minitab estimate the parameters from the data. If you like.14159. you can enter historical estimates for these parameters. there will be k-1 design variables and the reference level will be coded with all 0's. The distribution function you choose should depend on your data. Copyright © 2003–2005 Minitab Inc. date/time. Factor variables and reference levels You can enter numeric.Probit Analysis β c = unknown coefficient associated with the stress variable = natural response rate The distribution functions are outlined below: Distribution logistic normal extreme value (Euler constant) Here. all results − such as the percentiles − are based on these historical estimates. you can change this reference value in the Options subdialog box. pi in the Variance column of the table is 3. Minitab needs to assign one factor level to be the reference level. Distribution function Mean 0 0 Variance pi**2 / 3 1 pi**2 / 6 Estimating the model parameters Minitab uses a modified Newton-Raphson algorithm to estimate the model parameters. You would set the value when you have a historical estimate. Minitab obtains maximum likelihood estimates through an iterative process. change the maximum number of iterations for reaching convergence (the default is 20). or set the value. You want to choose a distribution function that results in a good fit to your data. or date/time factor levels. In this case. You can choose to estimate the natural response rate from the data. Probit analysis creates a set of design variables for the factor in the model. so you may want to specify what you think are good starting values for parameter estimates.

1795 109. they are displayed in a table in the Session window.0000 Probability 0. and cumulative failure probabilities. . The cumulative failure probabilities are the likelihood of failing rather than surviving. estimate survival probabilities for the stress values you specify.Estimate Stat > Reliability/Survival > Probit Analysis > Estimate You can estimate additional percentiles.9313 108.3982 109.6224 Upper 0.1281 Error 1. You can • • • estimate percentiles for the percents you specify.3715 1. These percentiles are added to the default table of percentiles.1504 95.5795 Upper 107. along with 95% fiducial confidence intervals. 50th. Table of Percentiles Standard Percent Percentile 1 2 3 4 104. we exposed light bulbs to various voltages and recorded whether or not the bulb burned out before 800 hours.2980 111. In this example. estimate cumulative failure probabilities for the stress values you specify.7692 Lower 0. Then we requested a survival probability for light bulbs subjected to 117 volts: Table of Percentiles 95. In this case.0% Fiducial CI Lower 101. what is the probability that an insect survives? You are looking for survival probabilities − estimates of the proportion of units that survive at a certain stress level. also known in the life sciences as the ED 10.9273 104. You can also request: • • • additional percentiles to be added to the table normal approximation rather than fiducial confidence intervals a confidence level other than 95% The Percentile column contains the stress level required for the corresponding percent of the events to occur. The probit analysis automatically displays a table of percentiles in the Session window. Common percentiles used are the 10th. 1% of the bulbs burn out before 800 hours.1656 As shown in the first and second columns of the first row. Survival and cumulative probabilities What is the probability that a submarine hull will survive a given strength of shock? At a given pesticide application. All rights reserved. Probit Analysis . the probability of failing before 800 hours at 117 volts is 0.1598 110.Reliability and Survivial Analysis Percentiles At what stress level do half of the units fail? How much pesticide do you need to apply to kill 90% of the insects? You are looking for percentiles. 166 Copyright © 2003–2005 Minitab Inc. You can also change the method of estimation for the confidence intervals and the level of confidence. survival probabilities. at 104. When you request survival probabilities.1104 105.8825 The probability of a bulb lasting past 800 hours is 0.2661 1.2308.0% Fiducial CI Stress 117. you exposed light bulbs to various voltages and recorded whether or not the bulb burned out before 800 hours. and 90th percentiles.9931 106. ED 50 and ED 90 (ED = effective dose).9931 volts. In this example.1997 1.5144 106.7692 at 117 volts.

Confidence level: Enter a confidence level for all of the confidence intervals. change the confidence level for the 95% confidence interval − click Estimate. Click OK. click Graphs. Confidence intervals: Choose to use two-sided confidence intervals (the default) or just an upper or lower confidence interval. If you like. Show graphs of different variables or by levels: Choose to display the graphs overlaid on the same page or on separate pages. Confidence Intervals Fiducial: Choose to display fiducial (default) confidence intervals. Probit Analysis . Deviance residuals versus event probability: Check to display a plot of the deviance residuals versus the event probability. Use these plots to identify poorly fit observations. If you like. The default is 95%. Estimate probabilities for these stress values: Enter one or more stress values or columns of stress values. Estimate cumulative failure probabilities: Choose to estimate cumulative failure probabilities.Probit Analysis Dialog box items Estimate percentiles for these additional percents: Enter one or more percents or a column of percent values. To request survival probabilities 1 2 In the Probit Analysis main dialog box. Survival plot: Check to display a survival plot. enter a value. lines and confidence intervals are drawn for each level. Normal approximation: Choose to display confidence intervals obtained with a normal approximation. If the plot looks cluttered. you can turn off the confidence intervals in the Graphs subdialog box. Probability plots A probability plot displays the percentiles. Check Survival plot. To draw a survival plot 1 2 3 4 In the Probit Analysis dialog box. Pearson residuals versus event probability: Check to display a plot of the Pearson residuals versus the event probability. All rights reserved. and cumulative failure plots. Copyright © 2003–2005 Minitab Inc. enter one or more stress values or columns of stress values. survival. survival. In Estimate probabilities for these stress values.Graphs Stat > Reliability/Survival > Probit Analysis > Graphs You can draw probability. Click OK. the closer the points fall to the fitted line. You can use the probability plot to assess whether a particular distribution fits your data. and cumulative failure plots. Estimate survival probabilities: Choose to estimate survival probabilities. Click OK. Dialog box items Probability Plot: Check to display a probability plot. You can also plot the Pearson or deviance residuals versus the event probability. Display confidence intervals on above plots: Check to display confidence intervals on the probability. In Confidence level. click Estimate. turn off the 95% confidence interval − uncheck Display confidence intervals on above plots. In general. 167 . Cumulative failure plot: Check to display a cumulative failure plot. the better the fit. You can also change the confidence level for the 95% confidence by entering a new value in the Estimate subdialog box. When you have more than one factor level.

the command terminates. Maximum number of iterations: Enter a number to change the maximum number of iterations for reaching convergence (the default is 20). Probit Analysis . Number of groups: Enter the number of groups for the Hosmer-Lemeshow goodness-of-fit test. − • To specify the maximum number of iterations. Natural response rate Estimate from data: Check to estimate the natural response rate.Reliability and Survivial Analysis Survival plots Survival plots display the survival probabilities versus stress. you must choose the option to display all results in the Probit Analysis . Dialog box items Estimate model parameters: Choose to estimate the model parameters from the data (the default). enter a positive integer. If you do not enter a value. Note Do not enter a starting value for the natural response rate here. . choose Use historical estimates and enter one starting value for each coefficient in the regression table. Minitab obtains maximum likelihood estimates through an iterative process. this value will be used as a starting value for the algorithm for estimating the natural response rate. define the event and reference factor level.Results dialog box. choose Estimate model parameters. The column should include one starting value for each coefficient in the regression table. Do one of the following.Results Stat > Reliability/Survival > Probit Analysis > Results You can control the display of Session window output. If Estimate from data is checked. the highest value in the column is used. Probit Analysis . 168 Copyright © 2003–2005 Minitab Inc. The column should include one value for each coefficient in the regression table. you can enter a value to define the occurrence of a success. Values should be listed in the order that they appear in the regression table. enter one starting value for each coefficient in the regression table. Reference factor level: Enter a reference level for the factor. Enter the values in the order that they appear in the regression table. If the maximum number of iterations is reached before convergence. Each point on the plot represents the proportion of units surviving at a stress level. the lowest value in the column is used. − To enter starting estimates for the parameters: In Use starting estimates. All rights reserved. click Options. To control estimation of the parameters 1 2 In the Probit Analysis main dialog box. which provide reasonable values for the "true" survival function. this value will be used as the natural response rate. Values should be listed in the order that they appear in the regression table. Enter the values in the order that they appear in the regression table. Option for Hosmer-Lemeshow Test: To see the results of this test.Options Stat > Reliability/Survival > Probit Analysis > Options You can choose to estimate model parameters and the natural response rate from the data or enter values. If Estimate from data is unchecked. and change the number of groups for the Hosmer-Lemeshow Test. Otherwise. Reference Options Event: If you have response/frequency data. Set value: Enter a value for the estimate of the natural response rate. To enter your own estimates for the model parameters. The survival curve is surrounded by two outer lines − the 95% confidence interval for the curve. this test bins the data into 10 groups. Use starting estimates: Enter a column that contains starting values for model parameters. By default. then click OK: • To estimate the model parameters from the data (the default). Use historical estimates: Enter a column that contains historical estimates for the model parameters.

Confidence limits for cumulative failure probabilities: Check to store the confidence limits for cumulative failure probabilities. This changes the confidence level for all confidence intervals. Cumulative failure probabilities: Check to store the cumulative failure probabilities. Display nothing Choose to suppress display of the results. Natural response rate: Check to store the natural response rate. enter the percents or a column of percents. • Choose Normal approximation to request normal approximation rather than fiducial confidence intervals. Copyright © 2003–2005 Minitab Inc.Storage Stat > Reliability/Survival > Probit Analysis > Storage You can store the Pearson and deviance residuals. the regression table. In addition. Probit Analysis . the multiple degree of freedom test (if the model contains a factor with more than two levels). Variance/covariance matrix: Check to store the variance/covariance matrix for the estimated coefficients. Dialog box items Enter number of levels in factor variable: If all of the samples are stacked in one column. Residuals Choose to store one or both of the residual types below: Pearson Deviance Characteristics of Fitted Distribution Choose to store any of the fitted distribution characteristics below: Percentiles: Check to store the percentiles. Standard error of estimates: Check to store the standard error of the estimated coefficients. characteristic of distribution and the Hosmer-Lemeshow goodness-of-fit test: You must choose this option in order to see the results of the Hosmer-Lemeshow test. multiple DF test and 2 goodnessof-fit tests Choose to display information about the response. the test for equal slopes (if the model contains a factor). table of percentiles and survival probabilities: Choose to display the distribution parameter estimates. regression table. In addition. Survival probabilities: Check to store the survival probabilities. Do any of the following. To modify the table of percentiles 1 2 In the Probit Analysis main dialog box. enter a value. click Estimate. enter the number of levels the column of grouping indicators contains. table of percentiles and/or survival probabilities (if requested in Probit Analysis Estimate). 169 . Standard error of percentiles: Check to store the standard error of the percentiles. distribution parameter estimates. • Change the confidence level for the percentiles (default is 95%): In Confidence level. Estimated coefficients: Check to store the estimated coefficients.Probit Analysis Dialog box items Control the Display of Results Choose one of the options below to control display of Session window output. and information on the estimated equation. the log-likelihood. log-likelihood. and the Pearson and Deviance goodness-of-fit tests. test for equal slopes. then click OK: • In Estimate percentiles for these additional percents. Stress for probabilities: Check to store the stress for survival probabilities or cumulative failure probabilities. characteristics of the fitted distribution. Information on Estimated Equation Choose to store any of the information on the estimated equation below: Event probability: Check to store the event probability. Show log-likelihood for each iteration of algorithm: Check to display the log-likelihood at each iteration of the parameter estimation process. Standard error of natural response: Check to store the standard error of the natural response rate. Response information. Percents for percentiles: Check to store the percents associated with the percentiles. All rights reserved. Confidence limits for percentiles: Check to store the confidence interval limits for the percentiles. Confidence limits for survival probabilities: Check to store the confidence limits for survival probabilities.

In Number of successes.262 Test for equal slopes: Chi-Square = 0.611 Log-Likelihood = -214. Session Window Output: Probit Analysis: Blows. and 132 volts. Click Estimate. In Estimate probabilities for these stress values.49188 DF 7 7 P 0. and define a success as: The bulb fails before 800 hours. choose Weibull.928 Type = A 170 Copyright © 2003–2005 Minitab Inc.61 1. You subject the two bulbs to five stress levels within that range: 108.000 0. Example of a Probit Analysis Suppose you work for a light bulb manufacturer and have been asked to determine bulb life for two types of bulbs at typical household voltages. Uncheck Display confidence intervals on above plots. enter Trials.12 P 0. Choose Response in success/trial format. 126. Type Distribution: Weibull Response Information Variable Blows Trials Value Success Failure Total Count 192 308 500 Factor Information Factor Type Levels 2 Values A. 120.51617 2.213 Goodness-of-Fit Tests Method Pearson Deviance Chi-Square 2. Trials versus Volts.MTW. enter Volts. . In Factor (optional). In Number of trials. B Estimation Method: Maximum Likelihood Regression Table Standard Error 7.926 0. All rights reserved.64 12. 1 2 3 3 4 5 6 7 8 Open the worksheet LIGHTBUL.0190 20. The typical line voltage entering a house is 117 volts + 10% (or 105 to 129 volts). Click OK in each dialog box.179368 0 Z -12. then click OK. Click Graphs. 114.000 0.67326 1. In Stress (stimulus). enter Type.159832 Variable Constant Volts Type B Natural Response Coef -97.58695 0. enter 117. Choose Stat > Reliability/Survival > Probit Analysis.0192 0.258463 DF = 1 P-Value = 0. From Assumed distribution.Reliability and Survivial Analysis Log-likelihood for last iteration: Check to store the log-likelihood for the last iteration. enter Blows.

764984 0.476 109.134 Copyright © 2003–2005 Minitab Inc.159 Percent 1 2 3 4 5 6 7 8 9 10 20 30 40 50 60 70 80 90 91 92 93 94 95 96 97 98 99 Percentile 101.342 121.341 101.922 135.19095 1.11242 1.010 130.472 123.63546 1.501 109.737 118.700 119.454 130.269 Table of Percentiles 95.997402 1.358 Standard Error 1.141 104.023 137.0192 127.3842 125.455 131.043 107.832 128.989 134.0% Fiducial CI Lower Upper 0.587 136.901 108.9868 104.754 131.0% Normal CI Lower Upper 17.1384 124.323 133.728520 0.868 113.07534 1.191 127.735850 0.15225 1.0% Fiducial CI Lower Upper 96.753 111.704348 95.864 135.698 112.641 134.706 116.551 135.780679 0.522 Parameter Shape Scale Estimate 20.639 112.881 123.280 132.28539 1.959732 0.802 109.720 110.683 132.41713 1.069 124.977596 0.0192 126.857 135.802 132.110 113.01968 1.261 135.1384 23.206 136.830608 Type = B Tolerance Distribution Parameter Estimates Standard Error 1.573 103.500 136.523 126.23831 Table of Survival Probabilities 95.980 133.287 110.068 107.731 106.23483 1.092 131.878549 Stress 117 Probability 0.50897 1.Probit Analysis Tolerance Distribution Parameter Estimates Standard Error 1.761 23.58695 0.506 110.207 138.898619 0.722 Parameter Shape Scale Estimate 20.108 130.737413 95. 171 .58695 0.246 115.299 128.201 122.542 137.34490 1.439 134.536 115.830361 0.082 120.3842 127.872 113.298 133.752 114.416 113.0% Normal CI Lower Upper 17.714 128.717911 0.16159 1.765 133.848 132.198 140.04522 1.11771 0.993 134.018 134.946 106.960 126. All rights reserved.135 111.001 108.550 124.943441 0.790097 0.84244 1.527 105.243 137.330 132.121 119.050 128.372 125.

240 133.01129 1.723 117.114 134.156 110.0% Fiducial CI Lower Upper 96.773 134. All rights reserved.16901 Table of Survival Probabilities 95.631 133.806 127.936720 0.17218 1.059 106.125 130.885656 0.584 127.444 136.982380 1.86171 1.800867 Table of Relative Potency Factor: Type Relative Potency 0.643 111.719 130.309 129.767 130.211 126.697 113.765 110.659 118.028 136.698947 0.00678 Comparison A VS B 172 Copyright © 2003–2005 Minitab Inc.418 131.452 123.969 132.725223 0.344 122.797 105.65621 1.925 128.513 131.36626 1.745980 0.991080 95.434 130. .927 135.720 115.607 104.837 108.845 125.781440 0.09453 1.13711 0.028 119.854567 Stress 117 Probability 0.256 120.30652 1.Reliability and Survivial Analysis Table of Percentiles 95.822 133.910842 0.772 Percent 1 2 3 4 5 6 7 8 9 10 20 30 40 50 60 70 80 90 91 92 93 94 95 96 97 98 99 Percentile 100.203 124.102 121.029 118.045 107.803 121.109 132.25563 1.53027 1.984 125.012 105.239 103.090 138.939 131.975363 1.43857 1.292 135.172 133.132 Standard Error 1.666 111.21135 1.500 131.429 125.484 134.496 108.267 110.0399 103.180 114.957949 0.101 112.719 134.461 134.673 102.947 107.901010 0.917912 0.449 132.476 114.0% Fiducial CI Lower Upper 0.493 135.742 109.828 112.727988 0.901 133.533 109.0% Fiducial CI Lower Upper 0.413 112.104 137.794 132.972 123.792908 0.508 108.698766 0.04700 1.471 100.823 112.168 131.828 129.123 106.

what percentage of the bulbs lasts beyond 800 hours? Eight-three percent of the bulb A's and 80% of the bulb B's last beyond 800 hours. • • • • Copyright © 2003–2005 Minitab Inc. The null hypothesis is that the model fits the data adequately. which tests that the slopes associated with the factor levels are significantly different. At 117 volts. is it a significant predictor? − natural response rate − the probability that a unit fails without being exposed to any of the stress. Since the test for equal slopes is not significant (p-value = .611).Output The default output consists of: • • • the response information the factor information the regression table. which includes the estimated coefficients and their − standard errors. − Z-values and p-values. Probit Analysis .Probit Analysis Graph window output: Interpreting the results The goodness-of-fit tests (p-values = 0. All rights reserved. the parameter estimates for the distribution and their standard errors and 95% confidence intervals.85 volts. in other words.928) and the probability plot suggest that the Weibull distribution fits the data adequately. which evaluate how well the model fits the data. At what voltage do 50% of the bulbs fail before 800 hours? The table of percentiles shows you that 50% of bulb A's fail before 800 hours at 124. In this case. the test for equal slopes. two goodness-of-fit tests. Therefore. 50% of bulb B's fail before 800 hours at 123. The parameter estimates are transformations of the estimated coefficients in the regression table.96 volts. the log-likelihood from the last iteration of the algorithm.262). 0. the comparison of light bulbs will be similar regardless of the voltage level. the light bulbs A and B are not significantly different because the coefficient associated with type B is not significantly different than 0 (p-value = .926. 173 . The Z-test tests that the coefficient is significantly different than 0. the higher the p-value the better the model fits the data.

which helps you to assess whether the chosen distribution fits your data. and 95% fiducial confidence intervals.Reliability and Survivial Analysis • • • the table of percentiles. Suppose you want to compare how the amount of voltage affects two types of light bulbs. 174 Copyright © 2003–2005 Minitab Inc.66 volts (117 x . This means that light bulb 1 running at 117 volts would fail at approximately the same time as light bulb 2 running at 114. the probability plot. All rights reserved. To get this output. lognormal. the relative potency − compares the potency of a stress for two levels of a factor. standard errors. and the relative potency is . and choose a Weibull. .98. or loglogistic distribution. which includes the estimated percentiles.98). you must have a factor.

pp.References." Journal of the Royal Statistical Society 38. [13] W. John Wiley & Sons. pp. Nelson (1985).E. J. "Experimental Survival Curves for Interval-censored Data. Turnbull (1974). Kececioglu (1991). [14] S. Lindqvist (1998). John Wiley & Sons.A. Statistical Methods for the Reliability of Repairable Systems. Basu (2000). [12] J." Reliability Engineering and System Safety 60. Kalbfleisch and R. Meeker and L. Applied Life Data Analysis. Goodness-of-Fit Techniques. Nelson (2003). D'Agostino and M. Escobar (1998)." Applied Statistics 22. All rights reserved.B. 290-295. John Wiley & Sons. W.W. W. 140-146. Inc. John Wiley & Sons. Recurrent Events Data Analysis for Product Repairs. The Statistical Analysis of Failure Time Data. 86-91. pp. Nelson (1990). 175 . Lawless (1982)." Journal of Quality Technology 17.P. D.Q. ASA-SIAM Seires on Statistics and Applied Probability.A.H. Academic Press. Stephens (1986). "Weibull Analysis of Reliability Data with Few or No Failures. 345. 169-173. Prentice Hall. Vols 1 and 2. Numerical Methods for Unconstrained Optimization. J. W. Rigdon and A. Peto (1973). and Other Applications. Statistical Methods for Reliability Data. John Wiley & Sons.D. Reliability Engineering Handbook. pp. Statistical Models and Methods for Lifetime Data. R.T.W. pp. Inc.Reliability and Survival Analysis [1] [2] [3] [4] [5] [6] [7] [8] [9] R. "Nonparametric Estimation of a Survivorship Function with Doubly Censored Data. Marcel Dekker. Nelson (1982). Disease Recurrences. (1972). Copyright © 2003–2005 Minitab Inc. 3. Turnbull (1976).L.F. Censored and Truncated Data. Murray. "TTT-based tests for trend in repairable systems data. Accelerated Testing. [10] B. Ed. [11] B. [15] W." Journal of the American Statistical Association 69. Reliability and Survival References . Prentice (1980). John Wiley & Sons. W. "The Empirical Distribution Function with Arbitrarily Grouped. Kvaloy and B. 13-28.

.

..................................................................................................... 143 L Least squares estimates........................... 66 Distribution ID Plot (Stat menu)....... 26 Distribution ID Plot (Right Censoring) ............................................... 122 Estimation Test Plans .......... 64 Arrhenius transformation............ 39............. 144 Parametric distribution analysis ........................ 43....................................................................... 67........................ 131 F Failure censoring .................................... 177 .Index Index A Accelerated life test model......... 72 Distribution Overview Plot (Stat menu) .. 123.......... 72 Duane plot........................ 132 M-failure test plan .... 130 Nonparametric Growth Curve (Stat menu) .. 118...... 65 G Goodness-of-fit statistics.................................. 143 C Censored data ..................... 55. 12 Estimation Test Plans (Stat menu)...... 102 Nonparametric Distribution Analysis (Stat menu) .................. 146...... 53 Nonparametric Distribution Analysis (Stat menu) ..... 141 Anderson-Darling statistic ........ All rights reserved........................................................ 37..... 80 Copyright © 2003–2005 Minitab Inc................................... 78.............. 146 Repairable systems analysis .... 23................. 7................. 55... 163 Probit Analysis (Stat menu) ........... 58...... 88 Mean cumulative difference function plot ...................... 104 Nonparametric distribution analysis .................... 124.................... 64 Percentiles...... 44............... 24...................... 35 Parametric Distribution Analysis (Stat menu) ............... 43............ 31 Distribution Overview Plot (Right Censoring) ......................................... 146..... 92 R Regression with Life Data ............................. 9 Multiple failure modes....... 8 Demonstration Test Plans (Stat menu) . 93....... 117 Growth curves. 24... 7 Frequency column .................... 35 Parametric Distribution Analysis (Right Censoring) ............................ 151 Regression with Life Data (Stat menu) ........... 88 Least squares method............................................................................... 43....... 8 Distribution ID Plot (Arbitrary Censoring)................... 26........... 16 Accelerated Life Test Plans (Stat menu).................................... 88 Loge (power) transformation ..................... 10 Probability of passing the demonstration test plot......... 167 Probability Plot (Reliability/Survival) 44.............. 77 Parametric Distribution Analysis (Stat menu) ........ 133 Mean cumulative function plot .... 12 Event plot .................................................... 115 Cumulative failure plot ................................. 156............... 31............. 165 Nonparametric Distribution Analysis (Arbitrary Censoring) ....................... 130 P Parametric Distribution Analysis (Arbitrary Censoring) .......... 130 Parametric Growth Curve......... 79 Probit analysis................ 118..................................................................... 88 Maximum likelihood method.............. 115............................................... 130 Nonparametric Growth Curves......................... 53 Nonparametric Distribution Analysis (Right Censoring) ...... 91 Parametric distribution analysis...... 66 Distribution Overview Plot (Arbitrary Censoring) ............................. 125 E Estimation methods for reliability analysis ......................... 143 M Maximum likelihood estimates..................................................................... 91 D Demonstration Test Plans.............. 10 Probability Plot... 39......... 18 Accelerated Life Test Plans ....................... 24............................... 64 Growth curve data......... 118 H Hazard plots................ 88 Parametric growth curves... 38........ 104 Parametric distribution analysis ............ 118 Parametric Growth Curve (Stat menu) ................... 93.. 73....................... 163 N Natural response rate ............................... 39.................... 80................................ 8........................................................... 167 Probit Analysis......... 130 I Inverse temperature transformation .......... 105 Nonparametric analysis.............. 43........................................ 58................................................................ 16 Accelerated Life Testing ........................................................................ 103................... 141 Accelerated Life Testing (Stat menu) . 54........... 37................ 102 Nonparametric Growth Curve........ 31 Distribution Overview Plot (Stat menu) ............ 118 Pearson correlation coefficient ......... 151 Relation plot....... 105 Parametric analysis .................................. 92.................................... 26 Distribution ID Plot (Stat menu)......... 39........................... 79 Accelerated life testing............................ 39............................................................ 156.................... 38................ 77 Parametric Growth Curve ............................................... 24......................... 166 POP graph... 35............... 116..

. 8 Total time on test plot .............................. 107 Comparing survival plots .......................................................................... 25................................. 144 Parametric distribution analysis................................................ 126 Trend tests for parametric growth curves .... 94 Probit analysis ............... 16 Accelerated life test plans ............ 80.. 59......................................................... 80 Probit analysis ..... 45..... 166 178 Copyright © 2003–2005 Minitab Inc...................... .......... 37...... 94... 65 Survival plots......... 107 Nonparametric distribution analysis ....................................................................... 59............................................ 166 Accelerated life testing ...................... 8 Estimation test plans...... All rights reserved........................ 168 Survival probabilities . 120 U Unstacked data................... 106 Parametric distribution analysis........... 144................... 37.................. 12.............................. 65 S Stacked data .... 25................... 45........ 106...............................................................Reliability and Survivial Analysis T Test plans .............. 12 Time censoring ........... 8.................................. 16 Demonstration test plans ....

You're Reading a Free Preview

Download
scribd
/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->