Monday, 7 January 2013: 11:30 AM
Room 6A (Austin Convention Center)
Utility companies in New England were subject to two severe weather events in the latter half of 2011. The aim of this research is to develop a tool to aid in predicting the extent of future damages from weather events in Connecticut and Western Massachusetts relating to the infrastructure of Connecticut Light & Power (CL&P) and its sister company, Western Massachusetts Electric Company (WMECO). Using a historic record of 50 storm events that caused major damages in the distribution network of both utilities, this model uses a restrictive framework by combining a logistic and gamma regression to model the probability of occurrence and probability density function of the number of damages to overheard distribution lines. The inputs to the model are weather parameters from high-resolution weather analysis by Weather Research and Forecasting (WRF) model initialized by GFS analysis fields, electric distribution infrastructure data, circuit tree-trimming history and land cover data. The model output is interruption rates (trouble spots per linear mile) of each circuit, delineated by circuit type (backbone or lateral) and damage type (primary or service) and aggregated by town in the CL&P and WMECO service territories. In this study we will present the model calibration parameters determined based on the historic events and evaluation of the model performance characteristics using different error metrics.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner