 |
Related Topics: |  |
 |
 |
|
 |
 |
 |
This topic provides a brief overview of the major analysis, data management and reporting capabilities provided by Weibull++.
You can also review an introduction to the Synthesis Platform and a list of what's new in Weibull++ Version 8.
ReliaSoft's Weibull++ provides the most comprehensive toolset available for reliability life data analysis, calculated results, plots and reporting. The software is also packed with tools for related analyses, such as warranty data analysis, degradation data analysis, non-parametric data analysis, recurrent event data analysis (for repairable systems) and reliability test planning.
The Weibull++ standard folio supports all life data types and all major lifetime distributions. You can analyze time-to-failure (complete), right censored (suspension), left censored, interval censored or free-form data, entered individually or in groups. Available lifetime distributions and analysis methods include:
1, 2 and 3 parameter Weibull
1 and 2 parameter Exponential
Normal and Lognormal
Gamma and Generalized Gamma
Logistic and Loglogistic
Gumbel
Bayesian-Weibull (which considers prior knowledge of the Weibull shape parameter)
2, 3 and 4 subpopulation Mixed Weibull (for situations when there are different trends in the data but you cannot identify a distinct failure mode for each data point)
Competing Failure Modes (CFM) analysis (which analyzes failure modes separately and then assumes a series reliability model in which each failure mode "competes" to cause the failure)
If you are not sure which model is appropriate for a given data set, the convenient Distribution Wizard automatically performs several types of goodness-of-fit tests in order to rank the available distributions.
Parameter Estimation options for standard life data analysis include Maximum Likelihood Estimation (MLE), Rank Regression on X (RRX) or Rank Regression on Y (RRY).
Weibull++ also provides Confidence Bounds for parameters, calculated results and plots. Depending on the specific analysis method used, the confidence bounds may be calculated using the Fisher Matrix, Likelihood Ratio, Beta Binomial or Bayesian approach.
The software provides a complete array of Calculated Results and Plots based on the analysis. For life data analysis, this includes:
Reliability or probability of failure
Reliable life (i.e., time for a given reliability, also called "warranty time")
BX% life (i.e., time for a given unreliability)
Mean life
Failure rate
Probability plots and pdf plots
Contour plots
Failure/suspension histograms, pie charts or timelines
The Weibull++ degradation analysis folio allows you to extrapolate the expected failure times of a product based on measurements that reflect how some performance measure (e.g., increase in crack propagation, decrease in tread depth, increase in vibration, etc.) has degraded for sample units over a period of time. The software offers a choice of the Linear, Exponential, Power, Logarithmic, Gompertz or Lloyd-Lipow models to analyze the degradation data, and generates Degradation vs. Time plots on either a linear or logarithmic scale.
Now in Version 8, Weibull++ also automatically calculates the life data model based on the extrapolated failure times and allows you to obtain calculated results and plots directly within the same folio — no need to transfer or link to another analysis folio!
Weibull++'s non-parametric LDA folio offers a choice of three methods for analyzing life data without assuming an underlying life distribution: Kaplan-Meier, Simple Actuarial and Standard Actuarial. This folio may be useful when dealing with unknown failure modes, when there is not enough data to assume a life distribution or when the data set does not fit any life distribution in a satisfactory way. Now in Version 8, Weibull++ also performs a parametric analysis directly within the same folio using the unreliability estimates that are generated by the non-parametric analysis.
Weibull++'s popular warranty analysis folio converts warranty claims data (sales and returns) that are readily available in many organizations into failure/suspension data sets that can be analyzed with traditional life data analysis methods. You can use this analysis to better understand the failure behavior of products in the field and to generate forecasts of future returns that will be covered under warranty. The software provides a choice of data entry formats to fit your particular needs: Nevada Chart, Times-to-Failure, Dates of Failure or Usage. The folio provides all of the special options you need to analyze the data in a way that’s appropriate for the available data and your organization’s warranty fulfillment practices. For example:
The Use Subsets option allows you to deal with non-homogeneous populations by analyzing data from different design iterations simultaneously and performing forecasts based on mixed sales data.
The Suspend After option allows you to take into account the possibility that failure data were not collected beyond the specified warranty period and/or to exclude predicted failures that will not be covered under warranty.
The Statistical Process Control feature (available for Nevada chart folios) can automatically detect abnormal sales or return periods and color-code the results to highlight specific data points you may wish to investigate further.
The Usage format allows you to enter returns data in terms of the amount of usage accumulated (e.g., mileage, cycles, etc.) rather than time in service. In Version 8, we have enhanced this feature by providing additional tools to help you configure the analysis to appropriately estimate the likely usage for units that are still operating in the field (i.e., suspensions).
Weibull++ provides a specialized folio designed specifically to capture system failure and repair data in an event log format like those commonly used in the machine tools and other industries. If you have a log that records the date/time when a system downing event occurred and the date/time when the system was restored to operation, the software converts this information to time-to-failure and time-to-repair data that can be analyzed with life data analysis techniques. Some useful options in this folio include the ability to:
Define the shift pattern that describes the time periods when the systems are operating each day.
Analyze data for a single system or multiple systems simultaneously.
Categorize events as "failures" or "events," and then specify whether non-failure events will be considered in the life data analysis.
Specify which component/assembly is responsible for each event, and then perform the analysis for any level of the system configuration.
The failure behavior of repairable systems is dependent on the history of repairs, and therefore traditional life data analysis methods (which treat failure data as independent and identically distributed) are usually not applicable. To provide the appropriate analysis treatment for such data, Weibull++ offers two analysis folios that can be used to analyze recurrent event data from repairable systems.
The Non-Parametric RDA folio uses the well-known Mean Cumulative Function (MCF) to plot the average number of recurring failures over a given period of time. The plot can be used to evaluate whether the number of failures is increasing or decreasing over time, to predict the future number of failures and to compare data sets from different designs, operating conditions or production periods.
The Parametric RDA folio uses the General Renewal Process (GRP) model, which takes into account the effectiveness of repairs on the condition of the system and models the cumulative number of failures over time. You can use the analysis to generate a variety of plots and calculated results, including the number of failures, failure intensity, mean time between failures (MTBF) and conditional reliability.
Version 8 offers a new Test Design Assistant that helps you select which reliability test design tools will meet your specific needs.
The Reliability Demonstration Test Design tool has been completely redesigned and expanded in Version 8. You can use the Parametric Binomial, Non-Parametric Binomial, Exponential Chi-Squared or Non-Parametric Bayesian methods to help choose the right test time/sample size for a reliability demonstration test. These methods can be used for designing a zero-failure test (where the reliability target is demonstrated if you don’t observe any failures during the test) and for tests with other quantities of "allowable failures" (e.g., one-failure test, two-failure test, etc.)
The Expected Failure Times Plot provides a visual depiction of the failure times you can expect to observe when you implement a particular test plan. If you perform the test and enter the actual failures as they are observed, you can use the plot to monitor whether the test is proceeding as expected or receive an early warning that adjustments may be needed.
The Difference Detection Matrix calculates how much test time is required before it is possible to detect (and demonstrate) a statistically significant difference in the life of two product designs.
Choosing an optimal reliability goal involves deciding on important trade-offs. For example, higher reliability typically requires higher production costs, but higher reliability will typically also lead to lower warranty costs and higher market share. The new Target Reliability Tool generates multiple plots that will help you select a target reliability that will minimize cost, maximize profit and maximize the return on an investment that affects reliability.
© 1992-2013. ReliaSoft Corporation. ALL RIGHTS RESERVED.