In probabilistic design it is common practice to use statistical models such as normal, lognormal, and Weibull to describe random design factors. However these distribution function models deviate in the lower tail, i.e. percentiles below 1%. The deviation is nonconservative in that since it predicts life longer than observed.

A Statistical Distribution Function called Adaptive Distribution Function Model similar to Abelkis model was developed. It is compatible with the collected data, and it produces conservative designs at low tail ends. It is also relatively easy to use.

This content is only available via PDF.