Gibbs Sampling for Game-Theoretic Modeling of Private Network Upgrades with Distributed Generation

Renewable energy is increasingly being curtailed,due to at times oversupply or network constraints. Curtailment can be partially avoided by smart grid management, such as ANM, but network reinforcement constitutes the long-term solution. Since network upgrades are expensive, recent interest has focused on incentivising private investors into participating in network investments. In this paper, we study settings where a private line investor installs a transmission line, but also provides access to other generators that pay a transmission fee. This model can be formulated as a Stackelberg game. Crucially the interdependent generation capacities built by renewable investors affect the resulting curtailment and profitability of projects. Optimal capacities rely jointly on stochastic variables such as the wind resource at the location. In this paper we how how big data and machine learning techniques, such as MCMC and Gibbs sampling, can be used generate observations from historic data and simulate multiple future scenarios, enabling optimal decision making regarding renewable energy investments. We present a game-theoretic formulation of the investment decision, and apply our methodology to a real network upgrade problem in the UK.