The Fit Life by X platform provides the tools needed for accelerated life-testing analysis. Accelerated tests are routinely used in industry to provide failure-time information about products or components in a relatively short time frame. Common accelerating factors include temperature, voltage, pressure, and usage rate. Results are extrapolated to obtain time-to-failure estimates at lower, normal operating levels of the accelerating factors. These results are used to assess reliability, detect and correct failure modes, compare manufacturers, and certify components.
The Fit Life by X platform includes many commonly used transformations to model physical and chemical relationships between the event and the factor of interest. Examples include transformation using Arrhenius (Celsius, Fahrenheit, and Kelvin) relationship time-acceleration factors and Voltage-acceleration mechanisms. Linear, Log, Logit, Reciprocal, Square Root, Box-Cox, Location, Location and Scale, and Custom acceleration models are also included in this platform.
You can use the DOE > Accelerated Life Test Design platform to design accelerated life test experiments. See Accelerated Life Test Designs in the Design of Experiments Guide.
You can use the Reliability Life Testing and Reliability Demonstration calculators to choose the appropriate sample sizes for reliability studies. These calculators are found at DOE > Sample Size Explorers. See “Sample Size Explorers” in the Design of Experiments Guide.
Meeker and Escobar (1998, p. 495) offer the following strategy for analyzing accelerated lifetime data:
1. Examine the data graphically. One useful way to visualize the data is by examining a scatterplot of the time-to-failure variable versus the accelerating factor.
2. Fit distributions individually to the data at different levels of the accelerating factor. Repeat for different assumed distributions.
3. Fit an overall model with a plausible relationship between the time-to-failure variable and the accelerating factor.
4. Compare the model in Step 3 with the individual analyses in Step 2, assessing the lack of fit for the overall model.
5. Perform residual and various diagnostic analyses to verify model assumptions.
6. Assess the plausibility of the data to make inferences.