Risk Limiting Dispatch in congested networks

Increased penetration of renewables requires improved power dispatch processes to limit integration cost. Existing Monte Carlo sample based approaches are difficult to implement in practice and result in difficult to interpret solutions. Recently, a two-stage dispatch procedure called Risk Limiting Dispatch (RLD) was proposed, and analytical solution was derived for uncongested networks. In this paper we extend RLD to a network with multiple congested links. We show that the network size can be reduced tremendously by carefully considering the possible congestions in the network. The key insight is to use first stage forecast values of renewables to predict the likely real-time congestions. Once the real-time congestions are predicted, we derive an equivalent second stage problem with dimension one greater than the number of congested lines in the original network. The detailed dispatch algorithm is given and illustrated with example networks.