Wednesday, December 17, 2014

Measuring Safety

"Should Safety be Measured?" was the opening question by professor Erik Hollnagel from University of Southern Denmark at todays meeting on "Measuring Safety" at Herlev Hospital in Copenhagen, Denmark.

Eric Hollnagel went on the argue, that in what he calls "Safety I" we don't measure safety. We measure the absence of safety by counting things, which are easy to count. Such as the number of fatalities, the number of near miss events, the number of outstanding HAZOP items, the number of outstanding safety reivews, the number of first aids, the number of fires, the numbers of leaks, the number of explosions, etc. The thesis is that the numbers go down that safety improves. However, behind these numbers are events, which clearly indicate the absence of safety, so hence we are not really measuring safety, but just its absence. Maybe, that is why safety is not improved.

We have been doing this for almost 80 years. But have we improved safety? I would say: No! We are still having people killed almost every day at a chemical plant or refinery somewhere in the world. And there is management saying, that what gets measured gets managed. Usually we measure something in order to use the measurement for feedback - or in some cases feedforward - control. However, if we are safe the numbers we measure go to zero. Then what good are they in feedback control? This situation has led Erik Hollnagel to focus on the work being done and how it is being done. He calls this "Safety II".

As I see it, the idea appear to be, that in professional sports one analyze how the best performers perform, and learn from them. So why not adopt this approach for work in chemical plants and refineries? What do you thing?

Eric Hollnagel and Sidney Dekker does not argues, that we should abandon the things we do to create safe or at least safer plants, such as HAZOP, PHA, FTA, ETA etc. They just argue, that alone that has not improved safety over 80 years of using this approach. An within the last 10 years even some of the icon performers of the 1980's have failed, e.g. DuPont. So currently it is difficult to see who the best are, whom we can learn from.