How can I publish a model based on a reliability DOE analysis?
Watch this video and others like it on ReliaSoft.tv
Starting in Version 9, you can publish models from DOE++ based on the results of a reliability DOE (R-DOE) analysis. The models then become available for use in other Synthesis applications (e.g., BlockSim or Xfmea) to represent the reliability of a component or assembly. R-DOE is a special category of design of experiments (DOE) in which traditional experiment design types are combined with reliability analysis methods to investigate the effects of different factors on the product life.
To build an experiment design for R-DOE in the Synthesis version of the software, create a standard design folio and edit the properties of the response you wish to analyze by specifying that the Response Type is life data. Then, in the Censoring drop-down list, choose which censored data type(s) will be entered. In this example, suspensions will be included in the data set.
After you have entered the data and performed the R-DOE analysis, the results will include the shape parameter of the life distribution and a DOE model for predicting the characteristic life parameter for a given set of factor values. To publish a reliability model that includes both parameters, click the Publish to Model icon on the Publishing page of the design folio's control panel.
You will then be prompted to choose a method for estimating the characteristic life parameter. In this example, optimization will be used to find the factor settings that maximize the life parameter. (Using prediction would allow you to enter the applicable factor values directly.)
Next, you need to specify the settings for the optimization and create the optimal solution plot. Once the plot is created and the maximized parameter value has been estimated, you can add the parameter to the reliability model and publish it by clicking the Publish to Model button.
The Publish Model window will appear, displaying all of the model's parameters. In this example, eta (the characteristic life parameter of the Weibull distribution) was estimated using the optimized factor settings. Click OK to publish the model.