Modeling Weather-Related Uncertainties in Electricity Generation for Mid-Term Risk Management

• Activity Number: 76
• Type: Contributed
• Date/Time: Sunday, July 31, 2011 : 4:00 PM to 5:50 PM
• Sponsor: Business and Economic Statistics Section
• Author(s): Virginie Dordonnat*+ and Jerome Collet
• Companies: EDF R&D and EDF R&D
• Keywords: risk management ; electricity markets ; weather based electricity generation ; dependence ; simulation

Abstract

Electricite de France (EDF) is the major competitor in the French electricity market. The company produces and sells electricity to other market participants as well as to end-use customers and therefore manages uncertainties related to both electricity production and demand. We concentrate here on the physical hazards involved in electricity generation, mostly the weather-related ones (e.g. wind farms production, inflows in hydroelectric power plants). Physical risk management means simulating electricity generation hazards from tomorrow up to 5 years. In order to generate adequate simulations we need to deal with all these uncertainties, include any predictive information and anticipate future changes (such as climate change), but also consider dependence between hazards. A first method is to use historical data : working with the same period of data is a natural way to account for the implicit dependence between hazards. We present the second way that implies statistical modelling and focus on why it is useful for risk management compared to historical data and how we consider dependence in our models.