Temperature Driven Bayesian Probabilistic Modelling of Electricity Demand, Capacity, and Adequacy

The declining costs for various distributed energy resources such as solar and energy storage is driving an increase in the penetration level of these resources at the grid's edge. The electricity market operator must account for these changes to effectively plan the system's demand, supply, and adequacy for various scenarios. This paper proposes a simplified methodology to create a probabilistic model of demand and supply which can be used to model resource adequacy as a function of temperature. This adequacy model is then translated to describe adequacy by duration of need. This description can then inform the duration of service needed from limited energy storage resources to reduce the probability of load being unserved. We first use a Bayesian additive model to infer the relationship between demand and available capacity as function of temperature. We then calculate the probability for when demand will be greater than supply for each unit increment of temperature. This probability can be described as a binomial random variable of demand being greater than supply for that hour. Finally, we estimate the duration of need by approximating the sum of binomial random variables for the day. With this methodology, one can rapidly simulate various supply mixes by fuel type to understand its effects on the final duration of need.