Probabilistic prosumer node modeling for estimating planning parameters in distribution networks with renewable energy sources

With the increase in distributed generation, the demand-only nature of many secondary substation nodes in medium voltage networks is becoming a mix of temporally varying consumption and generation with significant stochastic components. Traditional planning, however, has often assumed that the maximum demands of all connected substations are fully coincident, and in cases where there is local generation, the conditions of maximum consumption and minimum generation, and maximum generation and minimum consumption are checked, again assuming unity coincidence. Statistical modelling is used in this paper to produce network solutions that optimize investment, running and interruption costs, assessed from a societal perspective. The decoupled utilization of expected consumption profiles and stochastic generation models enables a more detailed estimation of the driving parameters using the Monte Carlo simulation method. A planning algorithm that optimally places backup connections and three layers of switching has, for real-scale distribution networks, to make millions of iterations within iterations to form a solution, and therefore cannot computationally afford millions of parallel load flows in each iteration. The interface that decouples the full statistical modelling of the combinatorial challenge of prosumer nodes with such a planning algorithm is the main offering of this paper.