Several Challenges of Predicting Utility Outages: A Description of Two "Tail End" Events

For virtually every electricity-providing utility company, unexpected outages are commonplace. However, damaging weather events can produce significant strain on a utility company by producing many outages in a very short amount of time. A wide variety of weather phenomena are capable of causing significant damage – including, but not limited to, severe thunderstorms, snowstorms, strong wind events, freezing rain events, and many others. If a utility company is better able to predict the number of upcoming outages in advance, they theoretically will be better prepared to deal with those outages in a timely (and cost-saving) manner. In practice, predicting these outages is a difficult endeavor. Central Hudson Gas & Electric has funded a research study with scientists at UAlbany, through their Center of Excellence, to develop an automated outage prediction system. This presentation will discuss various challenges associated with building an outage prediction model. An unexpectedly destructive March snowstorm and a damaging May severe weather event will be used to highlight some of these challenges. Additionally, data provided by the New York State Mesonet will be used extensively in the description of these events.