transformers, more than meets the eye

Earlier in the week, Yili Hong, a PhD student from Iowa State, gave an applied statistics seminar in our department about predicting transformer lifetimes. Our power system depends on having transformers up and working. Being able to predict transformer lifetimes (both individually and collectively) is useful for forecasting maintenance and installation. When looking at the data, it was clear that there are different types of transformers, and these classifications were important for making meaningful predictions. Although I didn’t understand some of the statistical techniques and bootstrapping algorithms applied, not being a statistician, I found a few things very interesting.  First, there are transformers that were installed in the 1930’s still working in the United States!  The old transformers have longer anticipated lifetimes than the new ones.  (Their lifetimes were modeled using a Weibull distribution with beta=2 vs. beta=6. Statisticians, discuss amongst yourselves).  This is incredibly depressing. We live in a world where everything is disposable. Even our necessary infrastructure.